> ## Content Index
> Fetch the complete content index at: https://www.rivativ.ai/llms.txt
> Use this file to discover other available public pages before exploring further.

# Predictive Gamma: July 2026
- URL: https://www.rivativ.ai/monthly-predictive-2026-07/
- Published: 2026-09-09T10:01:16.000Z
- Updated: 2026-09-09T10:01:16.000Z
- Author: Theo Paraskevopoulos

> The Predictive Gamma strategy returned +1.64% in July — a solid absolute result — but trailed the always-on straddle benchmark by 188 basis points as an unusually high proportion of sessions that the model passed on resolved profitably for indiscriminate sellers. The underperformance is the direct, expected cost of selectivity in a month where elevated geopolitical risk premium did not ultimately deliver the intraday tail moves it implied. Advisers should contextualise this result accordingly: the strategy's value lies in the sessions it avoids over a full cycle, not in matching an always-on baseline in any single month.

Predictive Gamma returned +1.64% in July, a strong absolute result in isolation, but the benchmark for this strategy — the return from selling a 0DTE S&P 500 straddle on every session — delivered +3.52%, leaving the strategy 188 basis points behind for the month. The underperformance reflects the strategy's fundamental design: the ML classifier issues a daily Long, Short, or Flat signal and commits capital only when it identifies a statistically elevated probability of a volatility crush or breakout, deliberately sitting out sessions it judges as low-conviction. 

In July, the confluence of geopolitically-driven intraday moves — oil surging on Iran escalation, semiconductor stocks whipsawing on the Moonshot AI release, and the Fed decision generating sharp cross-asset repricing — produced an unusually large number of sessions with realised intraday moves that rewarded indiscriminate straddle selling, regardless of regime. In short, the "quiet day" filter that is the strategy's edge worked against it in a month where even the noisy days resolved cleanly for short-volatility positioning. 

This is a known and accepted cost of selectivity: the strategy avoids the sessions most likely to produce large losses, and in exchange, it will periodically miss outsized gains when macro volatility resolves more benignly than the model's inputs suggest. A single month of benchmark underperformance against an always-on baseline is not a signal of model failure — it is the expected outcome when geopolitical risk premium is elevated but does not ultimately deliver the tail moves it prices in.

### Talking points

- A positive absolute return of +1.64% masked a month where sitting selectively on the sidelines proved costly. The strategy's ML model filtered out a number of sessions that, in hindsight, would have been profitable for indiscriminate straddle sellers — because geopolitical and macro noise in July resolved without the large intraday moves the model's risk signals anticipated. This is the honest trade-off of a selective approach.
- Underperformance against the always-on benchmark in any single month is expected and by design. The strategy's edge is avoiding the sessions most likely to deliver outsized losses — not maximising return in every environment. End clients should evaluate it over a full market cycle, where the avoided blow-up sessions matter far more than any one month of foregone upside.
- July illustrated that geopolitical risk premium does not always deliver the volatility it prices. Oil surging 24%, a divisive Fed meeting, and a major AI shock created the conditions that would ordinarily justify caution — yet intraday S&P 500 moves were ultimately contained. The model's conservatism in that environment was rational; the benchmark's outperformance was the result of risk that happened not to crystallise.

![](https://storage.ghost.io/c/bf/48/bf480f38-3f03-400d-b1af-7b0188655ac7/content/images/2026/09/2026-07-31_strategy_performance_Strategy_4.png)