The tail risk was not overload. It was a few big reversals
I have been investigating the "tail risk" (the occasional, sharp negative returns) that shows up in my sat2 strategy.
Rethinking the “Tail Risk” in the sat2 Strategy
I have been investigating the “tail risk” (the occasional, sharp negative returns) that shows up in my sat2 strategy. My previous hypothesis was that these spikes were caused by having too many open positions simultaneously, essentially “maxing out” the portfolio. I tested several variations of position caps, but the results were clear: capping the number of positions does not solve the problem.
Why position caps failed
I tested various caps, including total position limits and specific currency cross limits, on the sat2 strategy. The results showed that these caps were ineffective at mitigating the worst-case scenarios.
| Variation | Monthly Return | Worst Day Loss | Drawdown (DD) |
|---|---|---|---|
| Baseline | +0.514% | -2.23% | -9.27% |
| cap2 (Total cap 2) | +0.441% | -2.23% | -9.3% |
| cap3r (3 positions + 0.5% risk) | +0.441% | -2.23% | -14.9% |
| In other words, the worst-case days remained identical regardless of the caps. The “tail” isn’t caused by being fully loaded with five positions; it is caused by a strong, correlated reversal against just two or three open positions. Since the caps did nothing to help and, in some cases, even worsened the drawdown, I have rejected this approach. |
The effective solution: BE1.0R
The most effective protection is already in place: the BE1.0R (Break-Even) logic deployed in v1.11.0. When I isolated the sat2 strategy with this logic, the improvements were significant.
- Monthly Return: Increased from +0.514% to +0.566%
- PF (Profit Factor): Improved from 1.37 to 1.44
- Drawdown: Reduced from -9.27% to -7.06% Crucially, the BE1.0R logic successfully compressed the worst-day losses. For instance, the loss on June 23, 2022, was completely eliminated (going from -2.23% to 0.00%), and the impact of the 2016-11-09 event was reduced from -1.20% to -0.88%.
Implementing G4: Smart risk reduction
Since position caps were a dead end, I focused on a new system called G4, which scales down new entries during market stress rather than touching existing positions. By testing this against M1 intraday data, I found that scaling down only new entries captures the majority of the risk-reduction benefits without requiring the complexity of managing partial position closures. This method actually maintains or slightly improves monthly returns compared to the baseline, as it avoids “buying the dip” during a crash. I have now implemented this as the “Equity Shock Guard” in Core System v1.6.0 (EA v1.12.0). It monitors the US500 five-day return; if it drops below -3.0%, the system automatically reduces new risk by 50% for the next five days. The system is now live. With this new guard in place, I have the option to increase the capital allocation (kcap) to boost monthly returns by 35% to 62%, depending on my risk appetite. The core metrics for the updated system are:
- Monthly Return: +0.82%
- PF: 1.66
- Drawdown: -7.7%
- MC (Monte Carlo pass rate): 97.6% While the sat2 strategy still remains the primary contributor to worst-case scenarios, the G4 guard provides a robust, low-risk way to handle these shocks without the need for complex and ultimately ineffective position limits.
How this connects
This verification builds on earlier ones (what failed before and what I tried this time, comparisons between approaches).