The Financial Statement Model - Retired for Now

Once I got my Stock Prediction based on Annual (10-K) and Quarterly (10-Q) statement model working, I just wasn't happy with the R-squared on it. And I didn't feel comfortable investing in the picks it made (based on predicted returns). 

The R-squared on quarterly was so low, that trying to consider stocks it predicted for a quarter-long buy hold was just not feasible.

The R-squared on annual was considerably higher. But even then, it was not high enough to justify a stock purchase for a year-long tie-up of investment money.

Frankly, the stocks it was picking looked horrendous in many respects. Falling Knives, despite efforts to contain those, dominated the list. Others had low liquidity (read my earlier post on the Liquidity Effect) - and Solvency was an issue on them. Buying stocks with low or no liquidity and practically insolvent, and trying to hold them even a quarter, no less a year, is absolutely stupid.

I did Ensemble these models. But it didn't change the picture for me. And remember, I have Macro data and Macro Interactives in this model!

The conclusion: 
Statements (fundamentals) are important - but not for picking stocks based on them necessarily. You would have to combine the fundamentals with other things. 

I kind of knew this already, based on things I had read. I guess I needed to use the effort as a proving ground to myself.

So - in the end - I shelved these models. I learned a TON and it was great doing them. It built me into an AI Powerhouse with solid fundamentals in Quant Finance, an thorough understanding of Data Science and ML/AI algorithms, statistics, beefed-up math skills, etc.

I will move on. 

Target Encoding & One-Hot Encoding

I had completely overlooked these concepts earlier on, and somehow not until just now - this late in the game - did they come up as I was reading some of the advanced sections of the Jansen Algorithmic Trading book.

When you run a model, there are always going to be things you do NOT want to include. For many reasons (circular correlation, redundancy, etc). Generally, you would not want to include symbol, sector and industry in your model because the symbol after all, is nothing more than an index in the data - and the symbol in and of itself should have NOTHING to do with predicting returns.

So we keep an Exclusion array of metrics we don't want included:
EXCLUDED_FEATURES = [
    # 'symbol', 
    'date', 'year', 'quarter', 'close', 'fwdreturn',  # identity + target
    # 'sector', 'industry',  optional: include via one-hot or embeddings if desired
...
]

Symbol, Sector and Industry were initially excluded because they are text strings, and models don't like text data (some won't run without only numerics, some give warnings, others unpredictable results if they tolerate it).

Then I read about target encodings for symbols - which allow you to KEEP symbols in the data, and something called "one-hot" encodings for things like sectors and industries.

What these encodings do, is they take categorical data - like sectors and industries - and they convert them to a numeric so that they will be considered by the model. This captures the association between category and target, sometimes boosting model power.

So - adding the code to do this wasn't terribly painful. The symbol was encoded as a single encoded target feature (symbol_enc) - and removed from the above-mentioned EXCLUDED_FEATURES list.  The code to one-hot encode the Sector and Industry was snapped in, and off we went...

The R-squared dropped from .11 to .06. Yikes...
🪓  Pruning all splits - train, val, test - to the pruned set of features (final_features).
Training pruned model...
FULL R² -- Train: 0.4989, Validation: 0.0776, Test: 0.0673
PRUNED R² -- Train: 0.2216, Validation: 0.0348, Test: 0.0496
Selected FULL model based on test R².
Previous best test R²: 0.1126
Current model test R²: 0.0673
Current model did NOT exceed best. Loading previous best model and features.
Final Model Test Metrics -- R²: 0.1126, RMSE: 0.2133, MAE: 0.1471


I quickly realized, that by one-hot encoding the Industry, we got so many feature category permutations of the industries, that it undoubtedly screwed up the model.

I decided to run again, and only one-hot encode the Sector. So now with this run, we have a target encoding for symbol, and a one-hot encoding for sector. There are not that many sectors, so this doesn't become too unwieldy.

But - here is what we got on our SHAP analysis after this change:
🧮 Feature Selection Overview:
✅  consumer_sentiment_×_earnings_yield                SHAP=0.04511
✅  dilutedEPS                                         SHAP=0.02566
✅  rev_grw_4q                                         SHAP=0.01191
✅  evToSales                                          SHAP=0.01181
✅  roe                                                SHAP=0.01071
✅  symbol_enc                                         SHAP=0.01055
✅  business_confidence_×_capexToRevenue               SHAP=0.00925
✅  treasury_spread_x_debt_to_equity                   SHAP=0.00867
✅  cpippi_marginsqueeze                               SHAP=0.00719
✅  earningsYield                                      SHAP=0.00703
✅  inflation_x_debt_to_equity                         SHAP=0.00701
⛔  ppi_x_capextorevenue                               SHAP=0.00628
⛔  operatingCashFlow_to_totalAssets                   SHAP=0.00610
⛔  momentum_1y                                        SHAP=0.00586
⛔  gross_margin                                       SHAP=0.00575
⛔  realized_vol_3m                                    SHAP=0.00543
⛔  long_term_debt_to_equity                           SHAP=0.00526
⛔  rev_grw_qoq                                        SHAP=0.00500
⛔  vix_×_debt_to_equity                               SHAP=0.00480
⛔  goodwill_to_totalAssets                            SHAP=0.00432
⛔  cpi_x_netprofitmargin                              SHAP=0.00371
⛔  incomeQuality                                      SHAP=0.00360
⛔  rev_grw_qoq_to_totalAssets                         SHAP=0.00341
⛔  netDebtToEBITDA                                    SHAP=0.00340
⛔  evToFreeCashFlow                                   SHAP=0.00304
⛔  debt_to_equity                                     SHAP=0.00303
⛔  salesGeneralAndAdministrativeToRevenue             SHAP=0.00288
⛔  vix_x_evtoebitda                                   SHAP=0.00273
⛔  rev_grw_pop_sector_z                               SHAP=0.00268
⛔  eps_grw_qoq                                        SHAP=0.00258
⛔  evToEBITDA                                         SHAP=0.00258
⛔  interestBurden                                     SHAP=0.00255
⛔  momentum_1y_sector_z                               SHAP=0.00252
⛔  evToEBITDA_sector_z                                SHAP=0.00237
⛔  asset_turnover                                     SHAP=0.00235
⛔  totalEquity                                        SHAP=0.00231
⛔  net_margin                                         SHAP=0.00228
⛔  workingCapital                                     SHAP=0.00218
⛔  cc_delinquency_rate_×_debt_ratio                   SHAP=0.00209
⛔  inventory_turnover                                 SHAP=0.00209
⛔  capexToRevenue                                     SHAP=0.00197
⛔  freeCashFlowPerShare                               SHAP=0.00191
⛔  daysOfInventoryOutstanding_sector_z                SHAP=0.00186
⛔  fcf_ps_grw_qoq                                     SHAP=0.00186
⛔  eps_grw_qoq_to_totalAssets                         SHAP=0.00182
⛔  daysOfInventoryOutstanding                         SHAP=0.00180
⛔  daysOfPayablesOutstanding                          SHAP=0.00179
⛔  capexToRevenue_sector_z                            SHAP=0.00179
⛔  operatingCashFlow                                  SHAP=0.00173
⛔  totalEquity_to_totalAssets                         SHAP=0.00163
⛔  stockBasedCompensationToRevenue                    SHAP=0.00159
⛔  cash_ratio                                         SHAP=0.00158
⛔  evToOperatingCashFlow                              SHAP=0.00156
⛔  debt_ratio                                         SHAP=0.00156
⛔  eps_grw_4q                                         SHAP=0.00155
⛔  receivables_turnover                               SHAP=0.00153
⛔  ocf_to_current_liabilities                         SHAP=0.00151
⛔  totalDebt                                          SHAP=0.00146
⛔  operating_margin                                   SHAP=0.00143
⛔  debt_ratio_sector_z                                SHAP=0.00143
⛔  workingCapital_to_totalAssets                      SHAP=0.00135
⛔  operatingReturnOnAssets                            SHAP=0.00130
⛔  daysOfSalesOutstanding                             SHAP=0.00129
⛔  ordinary_shares                                    SHAP=0.00128
⛔  roa                                                SHAP=0.00127
⛔  earnings_surprise                                  SHAP=0.00122
⛔  bookValue                                          SHAP=0.00118
⛔  totalLiabilities                                   SHAP=0.00114
⛔  fcf_ps_grw_qoq_to_totalAssets                      SHAP=0.00113
⛔  goodwill                                           SHAP=0.00112
⛔  fcf_ps_grw_4q                                      SHAP=0.00106
⛔  eps_grw_pop_sector_z                               SHAP=0.00100
⛔  freeCashFlow                                       SHAP=0.00090
⛔  current_ratio                                      SHAP=0.00090
⛔  ebitda_margin                                      SHAP=0.00087
⛔  totalRevenue                                       SHAP=0.00082
⛔  quick_ratio_sector_z                               SHAP=0.00078
⛔  free_cash_flow                                     SHAP=0.00073
⛔  ocf_to_total_liabilities                           SHAP=0.00062
⛔  sector_Energy                                      SHAP=0.00062
⛔  returnOnTangibleAssets                             SHAP=0.00059
⛔  avgDilutedShares                                   SHAP=0.00050
⛔  sector_Basic Materials                             SHAP=0.00034
⛔  sector_Technology                                  SHAP=0.00025
⛔  sector_Real Estate                                 SHAP=0.00025
⛔  sector_Financial Services                          SHAP=0.00019
⛔  sector_Industrials                                 SHAP=0.00011
⛔  sector_Consumer Cyclical                           SHAP=0.00010
⛔  sector_Communication Services                      SHAP=0.00009
⛔  sector_Healthcare                                  SHAP=0.00007
⛔  sector_Utilities                                   SHAP=0.00004
⛔  sector_Consumer Defensive                          SHAP=0.00002
⛔  fcf_ps_grw_pop                                     SHAP=0.00000
⛔  rev_grw_pop                                        SHAP=0.00000
⛔  eps_grw_pop                                        SHAP=0.00000
⛔  unemp_x_rev_grw                                    SHAP=0.00000
⛔  ocf_to_net_income                                  SHAP=0.00000
⛔  quick_ratio                                        SHAP=0.00000
⛔  current_ratio_sector_z                             SHAP=0.00000

The symbol target encoding? SHAP likes it. It moves the proverbial needle.

But when it comes to the sector and industry, the one-hot encodings, neither of those seem to be adding any real benefit.

Our R-squared was better than the one we had with one-hot encoded industries - but less than our previous and best runs with NO encodings as shown below.

Symbol Target-Encoded and Sector One-Hot Encoded:
🪓  Pruning all splits - train, val, test - to the pruned set of features (final_features).
Training pruned model...
FULL R² -- Train: 0.4708, Validation: 0.0757, Test: 0.0868
PRUNED R² -- Train: 0.3847, Validation: 0.0779, Test: 0.0486
Selected FULL model based on test R².
Previous best test R²: 0.1126
Current model test R²: 0.0868
Current model did NOT exceed best. Loading previous best model and features.
Final Model Test Metrics -- R²: 0.1126, RMSE: 0.2154, MAE: 0.1485
Feature importance summary:
  → Total features evaluated: 78

So - one last run, with only the Symbol Target Encoding (not Sector and Industry)
🪓  Pruning all splits - train, val, test - to the pruned set of features (final_features).
Training pruned model...
FULL R² -- Train: 0.5169, Validation: 0.0788, Test: 0.0463
PRUNED R² -- Train: 0.2876, Validation: 0.0456, Test: 0.0083
Selected FULL model based on test R².
Previous best test R²: 0.1126
Current model test R²: 0.0463
Current model did NOT exceed best. Loading previous best model and features.
Final Model Test Metrics -- R²: 0.1126, RMSE: 0.2154, MAE: 0.1479
Feature importance summary:
  → Total features evaluated: 78

And...back to no symbol, sector, industry
🪓  Pruning all splits - train, val, test - to the pruned set of features (final_features).
Training pruned model...
FULL R² -- Train: 0.4876, Validation: 0.1095, Test: 0.0892
PRUNED R² -- Train: 0.2870, Validation: 0.0713, Test: 0.0547
Selected FULL model based on test R².
Previous best test R²: 0.1126
Current model test R²: 0.0892
Current model did NOT exceed best. Loading previous best model and features.
Final Model Test Metrics -- R²: 0.1126, RMSE: 0.2188, MAE: 0.1486

Once again - no encoding at all on symbol, sector and industry...
🪓  Pruning all splits - train, val, test - to the pruned set of features (final_features).
Training pruned model...
FULL R² -- Train: 0.4985, Validation: 0.1051, Test: 0.0969
PRUNED R² -- Train: 0.2742, Validation: 0.0981, Test: 0.0869
Selected FULL model based on test R².
Previous best test R²: 0.1126
Current model test R²: 0.0969
Current model did NOT exceed best. Loading previous best model and features.
Final Model Test Metrics -- R²: 0.1126, RMSE: 0.2160, MAE: 0.1487
Feature importance summary:
  → Total features evaluated: 78

Conclusion: These encodings did not move the R-squared at all, and in fact moved it backwards. So I need to make sure that there is no benefit to doing this encoding and if not, I will leave those out as the model was running originally.

I Have More Data Now - Enough for an AI RNN LSTM Model?

I have a LOT more data now than I did before. And an advanced architecture to process it.

Should I consider an RNN?

I knew I couldn't really pull it off with the Annual data I had - because by the time you split the data for training, validation, and testing there isn't enough to feed the algorithm.  

But - Quarterly! I have a LOT of quarterly data now, many statements per symbol across quarter-dates. ~70K rows of data!!!

So let's try doing an LSTM....I wrote a standalone LSTM, using Keras. Just a few lines of code. 

One important note about this! 

Do NOT mix your data processing, and or your XGBoost code, with neural network code!!! ALWAYS create a brand new virtual environment for your neural RNN code, because if you choose Keras or the other competing frameworks, they will require specific versions of Python libraries that may conflict with your data processing and/or XGBoost libraries!

Now. With that important disclaimer, the small sample of code. We will highlight in blue since my blog tool apparently has no code block format.

# -------------------------
# Train/test split
# -------------------------
X_train, X_test, y_train, y_test = train_test_split(
    X, y, test_size=TEST_SIZE, random_state=RANDOM_STATE
)

# -------------------------
# Build LSTM model
# -------------------------
model = Sequential()
model.add(LSTM(32, input_shape=(SEQ_LEN, len(feature_cols)), return_sequences=False))
model.add(Dropout(0.2))
model.add(Dense(16, activation='relu'))
model.add(Dense(1))  # regression output

model.compile(optimizer='adam', loss='mse')

# Early stopping
es = EarlyStopping(monitor='val_loss', patience=5, restore_best_weights=True)

# -------------------------
# Train
# -------------------------
history = model.fit(
    X_train, y_train,
    validation_split=0.1,
    epochs=50,
    batch_size=16,
    callbacks=[es],
    verbose=1
)

# -------------------------
# Evaluate
# -------------------------
y_pred = model.predict(X_test).flatten()
r2 = r2_score(y_test, y_pred)
mae = mean_absolute_error(y_test, y_pred)
print(f"R²: {r2:.3f}, MAE: {mae:.3f}")

Well, how did it go?

The previous XGBoost r-squared value, was .11-.12 consistently. Now, we are getting .17-.19. This is a noticeable significant improvement!
 

AI / ML - Data Source Comparison with More Data

"To validate generalizability, I ran the same model architecture against two datasets: a limited Yahoo-based dataset and a deeper Stockdex/Macrotrends dataset. Not only did the model trained on Stockdex data achieve a higher R² score (benefiting from more years of data), but the feature importances and pillar scores remained largely stable between the two. This consistency reinforces the robustness of the feature selection process and confirms that the model is learning persistent financial patterns, not artifacts of a specific data source or time window."

Details:

I wanted to move to an LSTM model to predict stock returns, but I was fortunate and patience enough to really read-up and plan this, before just diving into the deep end of the pool.

I learned that the "true AI" models - Transformers, RNNs, et al (of which LSTM is a subclass), require more data. I didn't have anywhere enough data using Yahoo, which gives 4-5 years (at best) of data. And because I was calculating momentum, yoy growth and such, I would always lose one of the years (rows) right off the bat - a significant percentage of already-scarce data.

So, in digging around, I found a Python library called stockdex. It is architected to be able to use multiple data sets, but the default is macrotrends. 

But using this library and source this left several challenges:

1. No quarterly data in the Python API - although the website does have a "Format" drop down for Quarterly and Annual.
2. The data was pivoted from Yahoo data. Yahoo put columns as items (x), and the time periods as rows (y). This stockdex API downloaded it opposite. 
3. The stockdex had no "names" for the items. 

Ultimately, I decided to use this because it returned a LOT more years of data.

1. First, I put some code together to download the raw data (statements), and then "pivot" the data to match Yahoo's format. 
2. Then, I used a mapping approach to change the columns from Macrotrends to Yahoo - so that I didn't have to change my logic that parsed statements.
3. I did have to do a run-tweak on the Metrics and Ratios, and fix certain columns that were not coming in correctly.
4. Lastly, I ran the model - same one as Yahoo and was able to keep the model logic essentially unchanged. 

The training took a LOT longer on Stockdex. The combined train+val had 14,436 rows on it.
Here is what we got:
FULL R² -- Train: 0.8277, Validation: 0.3441, Test: 0.3537
PRUNED R² -- Train: 0.7714, Validation: 0.3146, Test: 0.3429
Selected FULL model based on test R².
Final Model Test Metrics -- R²: 0.3537, RMSE: 0.3315, MAE: 0.2282
Feature importance summary:
  → Total features evaluated: 79
  → Non-zero importance features: 75

The model running and scoring, it took a very very long time. Finally it came out with this Top 25 list.
Top 25 Stocks by Final adj_pillar_score:
     symbol  adj_pillar_score  improvement_bonus  pillar_score
1209    RUN          1.156474               0.05      1.106474
884     LZM          1.020226               0.05      0.970226
97     ARBK          1.018518               0.00      1.018518
277     CGC          1.009068               0.02      0.989068
262     CCM          0.982228               0.02      0.962228
821    KUKE          0.964131               0.00      0.964131
1415   TRIB          0.963591               0.02      0.943591
1473   UXIN          0.961206               0.05      0.911206
571    FWRD          0.957156               0.05      0.907156
859    LOCL          0.935929               0.00      0.935929
1069   OTLY          0.896289               0.00      0.896289
894     MBI          0.895565               0.05      0.845565
1159   QDEL          0.890248               0.05      0.840248
1039    ODV          0.861127               0.00      0.861127
1522    WBX          0.860391               0.00      0.860391
1578   ZEPP          0.856097               0.02      0.836097
860    LOGC          0.846546               0.05      0.796546
990     NIO          0.811563               0.02      0.791563
1428    TSE          0.775067               0.05      0.725067
930    MODV          0.773322               0.05      0.723322
817    KRNY          0.770282               0.05      0.720282
1545    WNC          0.767113               0.02      0.747113
65     ALUR          0.756362               0.00      0.756362
813    KPTI          0.749644               0.05      0.699644
1316   SRFM          0.743651               0.00      0.743651

Then I ran the smaller Yahoo model:
Training pruned model...
FULL R² -- Train: 0.8181, Validation: 0.3613, Test: 0.2503
PRUNED R² -- Train: 0.8310, Validation: 0.3765, Test: 0.2693
Selected PRUNED model based on test R².
Final Model Test Metrics -- R²: 0.2693, RMSE: 0.4091, MAE: 0.2606
Feature importance summary:
  → Total features evaluated: 30
  → Non-zero importance features: 30

And, the Top 25 report for that one looks like this:
Top 25 Stocks by Final adj_pillar_score:
     symbol  adj_pillar_score  improvement_bonus  pillar_score
907    MOGU          1.532545               0.05      1.482545
1233    SKE          1.345178               0.05      1.295178
1170   RPTX          1.334966               0.02      1.314966
419      DQ          1.305644               0.05      1.255644
1211    SES          1.280886               0.05      1.230886
702    IFRX          1.259426               0.00      1.259426
908    MOLN          1.244191               0.02      1.224191
1161   RLYB          1.237648               0.00      1.237648
176    BHVN          1.232199               0.05      1.182199
512    FEDU          1.218868               0.05      1.168868
977    NPWR          1.205679               0.00      1.205679
1533     YQ          1.204367               0.02      1.184367
11     ABUS          1.201539               0.00      1.201539
58     ALLK          1.192839               0.02      1.172839
1249    SMR          1.154462               0.00      1.154462
63     ALXO          1.148672               0.02      1.128672
1482    WBX          1.140147               0.00      1.140147
987    NUVB          1.139138               0.00      1.139138
1128     QS          1.130001               0.02      1.110001
864     LZM          1.098632               0.00      1.098632
16     ACHR          1.094872               0.02      1.074872
1176    RUN          1.059293               0.02      1.039293
758    JMIA          1.053711               0.00      1.053711
94     ARBK          1.049382               0.00      1.049382
1086   PHVS          1.039269               0.05      0.989269

Symbols that appear in both top 25 lists:

• RUN (Stockdex rank 1, Yahoo rank 23)

• LZM (Stockdex rank 2, Yahoo rank 21)

• ARBK (Stockdex rank 3, Yahoo rank 24)

• WBX (Stockdex rank 16, Yahoo rank 17)

interesting...

Comparing the top sector reports:

Side-by-side Overlap Analysis Approach

Sector	Symbol(s) (Overlap)	Stockdex Rank & Score	Yahoo Rank & Score	Notes
Basic Materials	LZM, ODV, MAGN	LZM #1 (1.0202), ODV #2 (0.8611), MAGN #4 (0.7287)	LZM #2 (1.0986), ODV #3 (0.9233), MAGN #5 (0.8681)	Close agreement; Yahoo scores higher overall
Communication Services	KUKE, FUBO	KUKE #1 (0.9641), FUBO #4 (0.5936)	KUKE #5 (0.5559), FUBO #4 (0.6544)	Generally consistent rank order
Consumer Cyclical	NIO, UXIN, LOGC	UXIN #1 (0.9612), LOGC #2 (0.8465), NIO #3 (0.8116)	NIO #5 (0.9276), SES #2 (1.2809) not in Stockdex	Partial overlap; Yahoo picks also include SES, MOGU
Consumer Defensive	YQ, OTLY, LOCL	LOCL #1 (0.9359), OTLY #2 (0.8963), YQ #4 (0.6482)	YQ #2 (1.2044), FEDU #1 (1.2189), LOCL missing	Some overlap, differences in top picks
Energy	DWSN, PBF	DWSN #2 (0.6201), PBF #3 (0.4305)	DWSN #1 (0.7613), PBF #3 (0.3556)	Rankings closely aligned
Financial Services	ARBK, MBI, KRNY	ARBK #1 (1.0185), MBI #2 (0.8956), KRNY #3 (0.7703)	ARBK #1 (1.0494), GREE #2 (0.9502) missing MBI, KRNY	Partial overlap
Healthcare	CGC, CCM, TRIB	CGC #1 (1.0091), CCM #2 (0.9822), TRIB #3 (0.9636)	RPTX #1 (1.3350), IFRX #2 (1.2594) missing CGC,etc	Different picks mostly
Industrials	FWRD, EVTL	FWRD #1 (0.9572), EVTL #4 (0.7314)	NPWR #1 (1.2057), EVTL #5 (0.9657)	Some overlap
Real Estate	OPAD, AIV	AIV #1 (0.7303), OPAD #2 (0.7286)	DOUG #1 (0.8116), OPAD #5 (0.5578)	Partial overlap
Technology	RUN, WBX, ZEPP	RUN #1 (1.1565), WBX #2 (0.8604), ZEPP #3 (0.8561)	WBX #2 (1.1401), RUN #3 (1.0593), ZEPP missing	Strong agreement
Utilities	AQN	AQN #1 (0.7382)	OKLO #1 (0.8269) no AQN	Different picks

So - this is excellent model validation, I think. We see some differences due to the amount of time-period data we have, but the results are not widly different. 

I think I can now use this data in LSTM perhaps. Or whatever my next steps turn out to be, because I may - before LSTM - try to do some earnings transcript parsing for these if it's possible.

AI / ML - Feature Engineering - Earnings Surprises

I really felt that news would be a great thing to add to the model. But the problem with news, is that news is recent, and this data I am using with XGBoost is historical time-series data. 

If you added news, what would you do - cram the values only into the most recent year?

I think if you go with stuff that is changing and close to real-time, you need to re-think the whole model including the type of model. Maybe news works better with a Transformer model or a LSTM Neural Network model than a predictive regression model.

So - I am running out of new things to add to my model to try and boost the predictability of it (increase the R-squared).

Then I came up with the idea of adding earnings hits, misses and meets. A quick consult with an LLM suggested using an earnings_surprise score, so that not only do we get the misses/meets/beats counts but also we capture the magnitude.  A great idea.

I implemented this, and lo and behold, the earnings_surprise score moves the needle. Substantially and consistently.

The best thing about this, is that the earnings_surprise score is symbol-specific, and so it is not some macro feature I have to figure out how to interact with the symbol data. 

AI / ML - Here is Why You Backtest

My model was working nicely. 

It scored stocks on a number of fronts (pillars).

It used Profitability. It used Solvency. It used Liquidity. It used Efficiency.

These are the "four horseman" of stock evaluation.

I added some of my own twists to the "grading formula", in the form of macro variables (consumer sentiment, business confidence, et al).  I also had some trend analysis, rewarding trends up and penalizing trends down. I rewarded (and penalized) profitability, cash flow, etc. I had scaling done correctly, too in order to ensure a "fair playing field", and also some sector normalization as well.

When I ran the model, using XGBoost to predict 1-year forward return, the stocks at the top of the report looked great when I spot-checked them against various sites that also grade out stocks. I felt good. The r-squared I was getting from XGBoost and a SHAP-pruned feature run was at academic levels (as high as .46 at one point).

As part of some final QA, I ran the resultant code through AI engines which praised the thoroughness, and slapped me on the back reassuring me that my model was on a par with, if not superior to, many academic models.

Then - someone asked me if this model has this been back-tested. 
And the answer was no.  I had not back-tested it up to that point. I didn't think I was ready for back-testing. 

Maybe back-testing is an iterative "continual improvement" process that should be done much earlier in the process, to ensure you don't go down the wrong road.  But I didn't do that.

So, I ran a back-test. And to my horror, the model was completely "upside down" in terms of stocks that would predict forward return.  The AI engines suggested I simply "flip the sign" on my score and invert them. But that didn't feel right. It felt like I was trying to force a score.  

So - the first thing we did, was evaluate the scoring. We looked at correlation between individual scoring pillars and forward return. Negative.

We then looked at correlation in more detail.

First, we calculated Pearson (row-level) and Spearman (rank-level). correlations.

They were negative.

Then, we calculated Average Fwd Return by Score Decile. Sure enough, there was a trend, but completely backwards from what one would expect. 

Quality stocks with scores of 9,8,7,6,5 had negative values that improved as the decile dropped, while the shaky stocks (0,1,2,3,4) had graduated positive values.

The interesting analysis, was a dump of the correlations of each individual pillar to fwd return. The strongest were Profitability and Valuation, followed by MacroBehavior (macroeconomic features) but these were not strong. And the correlations were slightly negative, a couple slightly above zero positive.

But - one was VERY interesting. A log1p correlation between the "final composite score" to forward return that was noticeable if not sizable - but negative.

We experimented with commenting out the penalties, so we could focus on "true metrics" (a flag was engineered in to turn these off which made it easy to test). Re-ran the model, STILL the correlations with forward return were negative.

Then - we decided to remove individual pillars. Didn't change a thing. STILL the correlations with forward return were negative.

Finally, after the AI ensured me - after reviewing the code - that there were no scoring errors, the only thing left to try, aside of shelving the model for lack of success in predicting forward return, was to in fact put a negative sign on the score to invert it and "flip the score".

I did this. And, while the companies that bubbled to the top were shaky on their fundamentals, I did see cases where Analyst Ratings on these stocks were above (and in some cases way above) the current stock price.  

So here is the evidence that we have a model that IS predicting forward return, in a real way.

So - in conclusion. Quality does NOT necessarily equate to forward return.

What does??? Well, nothing in those pillars individually. But - when you combined all of these metrics/features into a big pot, and send them to a sophisticated regression modeler, it does find a magical combination that can predict a relationship with forward return that is linear, and depending on whether you flip that line one way or another, you can theoretically gain - or lose - a return on your money.

Now, if we had put money into those "great stocks" at the top of that prior list, and then had to watch as we lost money, it would have been puzzling and frustrating. But - do we have the courage to put money into these less-than-stellar fundamental stocks to see if this model is right, and that we WILL get a positive forward return? 

I guess it takes some experimentation. Either a simulator, OR, put $X into the top ten and another $X into the bottom ten and see how the perform. Which is what I might be doing shortly.