Trading System Signals on 01-12-2026

Capstone Trading
1 day ago
2 min read

The three Nasdaq strategies in Stock Index Portfolio 37 that are not included in Stock Index Portfolio 18 were the primary drivers behind SI 37 finishing as the top-performing portfolio on the day. Even in the face of headline risk, the tape continues to behave as if lower interest rates are the base case, while implied volatility remains muted and the VIX stays low.

Why hypothetical signals may differ from live results

In a well-built and properly executed trading program, hypothetical trading system signals should generally align with live performance. When they do not, the variance is typically explained by a short list of real-world exceptions:

Incremental transaction costs beyond the $25 round turn we assume (slippage + commissions)
Post-facto historical data revisions by the data vendor
Technical disruptions (disconnects, platform interruptions, or automation issues)

These factors can help or hurt in any given period, and they tend to wash out over a large sample size, but not always cleanly month-to-month.

Data revisions can move performance in either direction

On rare occasions, a data provider will revise historical data after the fact, which can slightly alter a backtest and the signals it would have generated. We have seen:

Winning trades disappear from hypotheticals after a data update, putting live results ahead of the hypothetical track record.
Losing trades disappear from hypotheticals after a data update, putting live results behind the hypothetical track record.

Neither outcome is “good” or “bad” in isolation; it is simply a reminder that historical data is not always perfectly immutable.

Technical issues sometimes require manual synchronization

From time to time, automation can be interrupted—by connectivity issues, platform glitches, or execution routing disruptions. When that happens, positions may need to be manually synchronized at prices that differ from the intended signal price. As with data revisions, this can work for or against us, and the impact becomes clearer only over a longer observation window.

The practical takeaway is straightforward: some months will run better than hypotheticals, and some months will run worse. That variability is normal in live trading.

Execution timing matters more than most traders realize

One final point is worth emphasizing: execution delay primarily increases dispersion, not necessarily the mean. If a trading system signal is executed a few seconds late—especially within the first three seconds—the standard deviation of slippage tends to widen, even when the average slippage remains roughly unchanged over time. This is an important nuance because many short-term traders assume that anything less than ultra-low-latency, co-located execution guarantees materially worse fills. In practice, that is not consistently true: a slight delay can sometimes produce a better fill, but it also makes outcomes less predictable by increasing the spread between good and bad fills.

A note on unstable strategies

Poorly developed strategies can also introduce discrepancies—particularly strategies that change or “switch” signals after a refresh, even when the underlying data has not changed. This is a common issue among newer developers who have not yet internalized concepts like signal stability, state management, and robust bar-by-bar logic. I shared a clear example of this behavior in a YouTube video a couple of years ago, and it remains a useful illustration of what to avoid when designing systematic models.