Stop guessing. Start building.

4,889 backtests. Two independent engines. Then one more check most backtests skip: the clock. Larry Williams' COT positioning index is supposed to sharpen a strategy's edge. We tested that claim directly: a plain RSI(2) mean-reversion strategy on the E-mini S&P 500, gated by Williams' COT index across all three trader groups, swept across lookback length, filter mode and level band. One baseline plus 4,888 filter-on combinations. First the engine. Our independent Python rebuild matched the original TradeStation study on all 4,889 variants. Trade count exact, net profit within a dollar on every row. Then the clock. That match used the date the data file stamps each report with, which is earlier than the day the CFTC releases it. So the published grid counts every report only from the close of its release day. On release days, the featured setup (Small Speculators, avoid the 35 to 75 middle band, 24-week lookback) scores a CAR/MaxDD of 0.20 against the unfiltered baseline's 0.51. It is the cell whose 27-cell neighbourhood has the highest mean CAR/MaxDD in its region, and it was chosen after the fact on the full sample. Zero of the 4,888 filter variants beat the baseline's CAR/MaxDD. 92.6% beat its average trade. That gap is the real finding. Full verification method and the release-day timing rule: https://statoasis.com/overfit/research/larry-williams-cot-filter

I have just rewritten my StrategyQuant X review from scratch, and one line in the pricing table turned out to matter more than everything else in the product. StrategyQuant X Starter costs $1,290. It gives you the Builder, the Retester, the Improver and tick-precision backtesting. That reads like the whole platform. It is not. Starter excludes advanced robustness tests, the walk-forward Optimizer, and Custom Projects. Those three begin at Professional, $1,490 — two hundred dollars higher. Sit with that for a second. The cheap edition will generate thousands of trading strategies for you and give you no proper way to tell which of them are real. A strategy generator is a wind tunnel, not a racing car. Every F1 team has a tunnel; what separates them is correlation — knowing which tunnel results survive a real track, and binning the rest. Starter sells you the tunnel and withholds the track. So the recommendation is simple and it runs against my own interest: buy Professional, or buy nothing. The $200 you save is the most expensive $200 in this category. Four things you should weigh this against. I am a paid StrategyQuant affiliate, I appear as a named testimonial on their own pricing page, I have held an Ultimate licence since 2019, and my flagship course is built around the platform. All four sit in the second heading of the review rather than a footnote, because most of the first page of Google for this product is written by affiliates who never say so. Verified prices, twelve screenshots from my own licence, and the seven honest limits: https://statoasis.com/overfit/research/strategyquant-x-review-the-no-code-way-to-build-trading-strategies

The most widely recommended filter in retail trading finished last of eight in my test. "Only take the trade if volume is above average." You have read it in every tutorial. The logic sounds airtight: heavy volume means real participation, light volume means a fake-out. I put 80 filters on Larry Connors' RSI(2) strategy and ran 15,552 backtests across SPY from 1993 to 2026 and S&P 500 futures from 2007 to 2026. Eight volume filters, eight volatility filters, and all 64 pairs of one with the other. Each filter was scored against its own unfiltered control across 96 matched configurations, rather than on the one where it happens to look good. Volume above the 50-day average improved profit factor in 28.1% of those pairings. Volume BELOW 0.8 times average improved it in 71.9%. That is not a rounding difference. The advice is pointing the wrong way for this kind of entry. I do not think the confirmation logic is wrong so much as misapplied. It was built for breakouts, where participation is the thesis. When you buy a two-day washout you are betting the sellers ran out of people, and a heavy-volume bar says they did not. Same indicator. Opposite trade. Opposite reading. The filter that actually earned its place was not a volume filter at all. ATR percentage below its own 100-day median, which in plain terms means: only take the setup when the market has been calmer than usual. Profit factor up 0.52, better in 75% of pairings, worst drawdown down 5.6 percentage points, and it still lets half the signals through. It costs 1.60 percentage points of annual return. Every filter I tested cost something. Not one raised it. Full write-up, all sixteen filters, the 64-pair heatmap and the limits: https://statoasis.com/overfit/research/boost-your-rsi2-strategy-for-sp500-by-48-with-this-volume-filter