Works for any time-dependant prediction task • Finance, ML, Sales, Weather, Energy

Did Your Predictions
Actually Work?

Stop measuring how close you were. Start measuring if you would have won.

Upload Your Predictions

Drop one or more CSVs with predictions & actual outcomes (6 columns maximum)
Works for stocks, sales, weather, sports-anything

Download Example CSV
Download Example CSV

No account required. Anonymous access includes 5 successful tool runs per rolling week.

Your Results

Success Score
-
How useful these predictions actually are

Forecast vs Actuals

Actuals stay visually dominant, the strongest forecast stays emphasized, and weaker candidates remain visible without taking over the chart.

Forecast Comparison

What To Do Next

Move from evaluation into advisory or execution.

Next layer

Decision Intelligence

Diagnose the dataset, select the target, assess dataset health, and get a structured model path before training.

Open Decision Intelligence
Execution layer

AutoML

Carry forward the recommended shortlist, benchmark concrete candidates, and keep execution aligned with the upstream decision flow.

Open AutoML

Segmented File Results

Direction Accuracy
-
Did you predict up/down correctly?
Profit Score
-
If you acted on these, would you win?
CER
-
Balance of usefulness vs accuracy
Risk Score
-
Reward relative to downside risk
Worst Case
-
Biggest mistake in your predictions

Interpretable Error Metrics

MAPE
-
Mean absolute percentage error
MAE
-
Mean absolute error in original units
MAE %
-
MAE relative to mean actual magnitude
Coverage ±10%
-
Predictions within 10% of actuals
Bias
-
Systematic over or under prediction
Max Error
-
Worst single-point prediction miss

What's the Success Score?

Think of it like this: If you bet money on your predictions, would you make profit?

Why Not Just "Accuracy"?

Being close isn't enough. Wrong direction can lose money.

Who Uses This?

Anyone who needs predictions that work, not just look good on paper.

Real Examples That Make Sense

This works for any prediction, not just finance

Stock Trading

Scenario: You predict a stock at $101, it goes to $99.
Traditional: "Close! Only 2% error."
Quantsynth: "Wrong direction = you lost money."

Weather Forecasting

Scenario: You predict 75°F, actual is 73°F.
Traditional: "Good job!"
Quantsynth: "Did people bring jackets? That's what matters."

Sports Betting

Scenario: You predict a 10-point win, they win by 3.
Traditional: "Pretty close!"
Quantsynth: "You predicted a win and they won = profit."

Sales Forecasting

Scenario: Predict $100K sales, actual is $80K.
Traditional: "20% error"
Quantsynth: "Did you overstaff? That costs real money."

ML Model Selection

Scenario: Model A: 95% accurate. Model B: 88% accurate.
Traditional: "Pick A"
Quantsynth: "Which one makes fewer expensive mistakes?"

Real Estate

Scenario: Predict price increase, it decreases 5%.
Traditional: "Small error"
Quantsynth: "Wrong direction = you bought at the peak."

Why Traditional Metrics Miss The Point

Comparing what matters vs what gets measured

WHAT MATTERS TRADITIONAL APPROACH QUANTSYNTH APPROACH
Getting Direction Right Completely ignored (only cares about distance) Primary focus (80% of success)
Actual Success Rate No connection to outcomes Directly measures if it works
When You're Wrong Big mistakes kill your score Understands asymmetric risk
Confidence Levels Treats all predictions equally Rewards being confident when right
Real-World Use Great metrics, bad decisions Optimizes for actual utility

Metrics That Reflect Real Outcomes

Not how close your predictions were, but how useful they actually are

Forecast Investment Score (FIS)

FIS is QuantSynth's proprietary metric for evaluating forecast behavior from a decision-oriented perspective. It complements, but does not replace, conventional error metrics.

  • Compare candidates under the same dataset and validation design
  • Investigate disagreements with MAE, MSE, RMSE, and MASE
  • Confirm rankings on held-out or walk-forward data
How to read it:
Higher values generally indicate a more favorable result under the FIS framework. They do not represent expected return, success probability, or a performance guarantee.

Confidence–Efficiency Ratio (CER)

CER is QuantSynth's complementary proprietary metric for comparing forecasts from predictive and decision-oriented perspectives. It can rank candidates differently from FIS and conventional error metrics.

  • Use it as an additional ranking perspective
  • Review its result beside traditional error diagnostics
  • Avoid isolated comparisons across unrelated datasets
How to read it:
Interpret CER comparatively within a consistent evaluation. A favorable CER ranking is not universal evidence that a model is superior.

How These Metrics Work Together

FIS and CER provide distinct proprietary comparison views. Use both alongside traditional errors, validation stability, baselines, and application-specific outcomes. Neither metric establishes deployment readiness on its own.

See The Difference in Action

How traditional metrics get fooled by common prediction patterns

The Lagged Predictor

This model just predicts yesterday's price. It's incredibly common (moving averages, momentum strategies) and can look favorable on traditional metrics because it remains close to recent values. In a timing-sensitive task, it may react too late.

Sample Predictions:
Day 1: Predict $100 → Actual $102
Day 2: Predict $102 → Actual $98
Day 3: Predict $98 → Actual $103
Day 4: Predict $103 → Actual $101
Why it fails:

By the time you get the signal, the move already happened. You're always entering after the relevant event. This is a hypothetical timing failure, not a trading-performance result.

METRIC COMPARISON

R² Score 0.92

"Excellent fit!"

MAE 1.8

"Very accurate!"

FIS Score 0.02

"Weak under this hypothetical decision rule"

Hypothetical interpretation: delayed signals could be unfavorable after costs. Validate with realistic execution assumptions.

Built for Professional Use

The transparency and privacy you need for sensitive data

Privacy First

Predictions processed in-memory. We evaluate your utility, we don't steal your alpha. No data storage, no tracking, no leaks.

Transparent Evaluation

Intended use, interpretation, evidence, and limitations are documented. The scoring mathematics and implementation remain proprietary.

Full Transparency

Download reports include all the necessary data to complete your analysis. No black boxes.

© 2026 – Francisco Cardoso. All rights reserved.

The content of this website, including design, code, models, text, graphics, and displayed results, is protected under applicable copyright and intellectual property laws. Unauthorized reproduction, modification, reverse engineering, systematic data extraction, or use for competitive purposes without prior written consent from the owner is strictly prohibited.