Ali Casey | StatOasis

Ali Casey | Quant research for traders who test before they trust. Every study published, including the ones that failed.

[Ali Casey | StatOasis @_StatOasis]

The honest frame: this is daily bars, tested mechanically, frictionless. ICT's intraday discretionary craft remains untested — and untestable from published rules.

Full tables, horizon curves, and the RSI irony:

https://t.co/la4jgkfMXt

[Ali Casey | StatOasis @_StatOasis]

I backtested ICT / Smart Money Concepts. Mechanically. 648 backtests across 4 markets.

Order blocks, fair value gaps, liquidity sweeps, Optimal Trade Entry — versus 3 boring entries, a coin flip, and buy-and-hold.

What survived: 🧵

[Ali Casey | StatOasis @_StatOasis]

Cross-market survival — the formal test. Does any family's best variant beat BOTH baselines on ALL four markets?

Survivors: none.

Order Block's 0.241 return per unit of drawdown on SPY fell to 0.104 on QQQ. Rankings reshuffle every market. That's what noise looks like, not a mechanism.

[Ali Casey | StatOasis @_StatOasis]

17,424 backtests, and here is what they cannot tell you.

Every figure in my Crabel rebuild is frictionless. No commission, no slippage, no spread.

The winning settings average a fraction of a per cent per trade. That is where your costs decide it.

https://t.co/uBIXo0zXAC

[Ali Casey | StatOasis @_StatOasis]

Every 4% dip triggers the same question: is this the big one?

History's answer, from all 127 completed S&P 500 declines of 3%+ since 1928:

• 73 reached −5%
• 26 reached −10%
• 12 became bears (−20%+)

Most dips die young. But once one hits −10%, it's nearly a coin flip (46%) it keeps going.

Full study: https://t.co/ZAqAWN5ysg

[Ali Casey | StatOasis @_StatOasis]

The best filter I tested made LESS money.

An OBV check on a MACD cross, 3,072 pairs:

Drawdown smaller: 85.7%
Profit factor better: 67.3%
Net profit better, long: 30.4%

Fewer trades. Fewer winners.

Judge a filter on profit and you bin the good ones.
https://t.co/PltuwdJS52

[Ali Casey | StatOasis @_StatOasis]

Decoded Dalio's chart, including the two columns nobody reproduces.

15 uncorrelated streams take your chance of a losing year from 40.1% to 16.6%.

The 1% at the bottom of his axis? That needs 100 streams.

Volatility falls with the square root of the count.

https://t.co/eJjuDmB759

[Ali Casey | StatOasis @_StatOasis]

Stock/bond correlation by decade:

1970s +0.25
1980s +0.31
1990s +0.27
2000s -0.33
2010s -0.45
2020s -0.08

The hedge that makes 60/40 work only showed up after 2000.

2022 wasn't an anomaly. It looked like the old normal.

https://t.co/oE2xQII3NZ

[Ali Casey | StatOasis @_StatOasis]

17,424 backtests on Toby Crabel's thrust patterns, eight markets.

Up thrust, 10-bar hold: +0.39%.
An average day in the same markets: +0.29%.

The pattern's own edge is 0.09 percentage points. Hold it 20 bars and you are behind, at -0.16.

https://t.co/uBIXo0zXAC

[Ali Casey | StatOasis @_StatOasis]

The OBV MACD is not a better MACD.

3,072 matched pairs. Same market, same periods, same side, same exit. Only the input changes.

Volume won 44.4% of them on profit factor.

Worse than a coin flip. Keep your MACD and let OBV veto it instead:
https://t.co/PltuwdJS52

[Ali Casey | StatOasis @_StatOasis]

The biggest finding is the one I didn't expect.

Stock/bond correlation by decade:
1970s +0.25
1980s +0.31
1990s +0.27
2000s -0.33
2010s -0.45

The 60/40 hedge is ~25 years old, not a law. 2022 looks like the old normal.

Full study:
https://t.co/oE2xQII3NZ

[Ali Casey | StatOasis @_StatOasis]

Ray Dalio's diversification argument: 15 good, uncorrelated return streams cut your risk ~80% without cutting your return.

The maths is real. I rebuilt it from the formula.

Then I checked the part nobody checks — whether the streams exist.

[Ali Casey | StatOasis @_StatOasis]

And the crisis story is backwards.

Across the 4 asset classes, correlation went NEGATIVE in five of six major declines since 1983 — including 2008 (-0.031) and COVID (-0.019).

Only 2022 went positive (+0.119).