Verax
Not financial advice

Learning centre

Quant without the hype

Quant is not magic. It's the discipline of not fooling yourself with data. This page teaches you to test ideas honestly — and recognise when you're not.

01

Learning path

0 / 6 steps marked done

What is a quant strategy?

A quant strategy is a fixed rule applied to data: "buy when X, exit when Y." No gut feel, no news, no hunches. The rule doesn't change between Monday and Friday. The payoff is consistency — the trap is thinking rules equal certainty.

What a backtest actually is

A backtest replays your rule on historical prices and asks: "if I had done this, what would have happened?" It's not a prediction. It's a historical simulation — the future will differ, often a lot. The value is in the questions it lets you ask, not the number it gives you.

Try it in Backtest Lab

Reading the scoreboard

Three numbers matter most. Return: how much did it grow? Drawdown: how bad was the worst drop? Sharpe ratio: return divided by risk — how smooth was the ride? A strategy with 15% return and 40% drawdown may be worse than one with 10% return and 12% drawdown.

See metrics in Backtest Lab

The #1 trap: overfitting

Every strategy looks great if you tweak it enough on past data. That's overfitting — memorising noise. The test is always: how does it do on data it has never seen? A rule that only works at one magic parameter value almost certainly found a coincidence, not an edge.

See it live below ↓

Honest testing protocol

Reserve a chunk of your data — never touch it during building. Build on the first part, test on the second. Check that nearby parameter values give similar results (sensitivity). Add realistic transaction costs. If it still works: you have weak evidence of an edge. Still weak — markets adapt.

Sensitivity in Backtest Lab

What quant can and cannot do

Quant can: remove emotion, enforce discipline, test ideas systematically, manage risk mechanically. Quant cannot: predict the future, eliminate loss, guarantee outperformance, make markets rational. Most professional quant funds underperform a simple index after fees. Tools, not oracles.

Try a single asset analysis
02

See overfitting happen

Move the slider — watch history improve, then watch it collapse

Strategy complexity — level 1

SimpleOverfit
IN-SAMPLE (building)Strategy 32.1%B&H -2.0%click "Test on unseen data"
Simple rule, modest in-sample result. Likely captures something real, even if imperfect.
03

Concepts in action

Two ideas you can feel, not just read

Why diversification works (and when it doesn't)

Drag the slider to change how correlated Asset A and B are. Watch how the combined portfolio's volatility changes — even when both assets have the same 20% annual volatility on their own.

Correlation: +0.8

−1 (perfect hedge)0 (independent)+1 (lockstep)
Asset A (alone)
20.0%
Asset B (alone)
20.0%
50 / 50 Portfolio
19.0%

Moderate correlation. Portfolio volatility is 5% lower than each asset alone — some benefit.

How the bot sizes positions

The bot targets a constant portfolio volatility by holding more of calm assets and less of wild ones. Drag the target to see positions change — no fundamental analysis, just math.

Target portfolio volatility: 15%

5% (conservative)30% (aggressive)
100%

Bond ETF

8% vol

83%

Stock ETF

18% vol

43%

Commodity

35% vol

Formula: position = target_vol ÷ asset_vol — capped at 100%. The bond ETF gets the largest allocation because it contributes the least risk per dollar invested.

04

Terminology

Every term in plain language

Returnbasics

Total gain or loss as a percentage of starting capital. A 20% return on $10 000 = $12 000.

Volatilityrisk

How wildly a price swings. Measured as standard deviation of daily returns. More swing = more risk.

See it in the tool →
Drawdownrisk

The drop from any peak to the subsequent trough before recovery. The pain you feel while holding. Closer to 0% is better.

See it in the tool →
Sharpe ratiorisk

Annualised return divided by annualised volatility. Measures smoothness of gains. A calmer 10% can beat a wild 15%.

See it in the tool →
Betabasics

How much the portfolio moves when the market moves. Beta 1.0 = moves with market. 0.5 = half as much.

Alphabasics

Return above what market exposure alone explains. True alpha is rare and erodes fast as it gets traded away.

Momentumstrategy

Assets that went up recently tend to keep going up (for a while). A persistent but fragile anomaly.

See it in the tool →
Mean reversionstrategy

The opposite of momentum: prices tend to drift back toward average after extreme moves. Works until it doesn't.

Buy & holdbasics

Buy once, hold forever, never react. The honest benchmark that beats most active strategies after fees.

See it in the tool →
Backtesttesting

Replaying a rule on historical data. Tells you what would have happened — not what will happen.

See it in the tool →
Out-of-sampletesting

Data not used during strategy development. Testing on it is the only credible performance check.

Overfittingtesting

Tuning a strategy until it looks perfect on past data. It then fails forward because it memorised coincidences, not patterns.

Diversificationrisk

Owning assets that don't all move together. When correlation is low, one can fall while another holds.

See it in the tool →
Correlationrisk

How similarly two assets move. 1.0 = lockstep. 0 = independent. Negative = opposite. Lower = better diversification.

Risk-adj. returnrisk

Return relative to risk taken. A lower return achieved with far less risk may be the better outcome.

See it in the tool →
Return distributionrisk

The histogram of how often each return size appeared historically. Reveals whether gains and losses are symmetric, skewed, or have fat tails.

See it in the tool →
Fat tailsrisk

Extreme events (crashes, spikes) happen far more often than a normal bell curve predicts. Real markets have fat tails — rare big moves are not as rare as textbooks say. High kurtosis = fatter tails.

See it in the tool →
Skewnessrisk

How asymmetric a return distribution is. Negative skew = rare but severe losses on the downside. Most assets have negative skew: small daily gains, occasional large drops.

See it in the tool →
Kurtosisrisk

How heavy the tails are relative to a normal distribution. Excess kurtosis > 0 means more extreme days than expected. Related to fat tails. Most stock indices have kurtosis above 4.

See it in the tool →
Drawdown recoveryrisk

The time it takes to return to the previous peak after a drawdown. Depth alone is misleading — a -30% drop recovering in 3 weeks is very different from one that takes 3 years.

See it in the tool →
Seasonalitystrategy

Historical tendencies for returns to be higher or lower in certain calendar months. Observed in data, but not reliably predictive — markets adapt once patterns become widely known.

See it in the tool →
05

Honest answers to common myths

Evidence first, no hype either way

Ready to test an idea?

Backtest Lab applies everything on this page: out-of-sample splits, sensitivity checks, realistic costs. No signals. No promises. Just the historical record, honestly presented.

Backtest Lab →Single Asset →

For education only. Not investment advice. Past results do not predict future performance.