Risk control through diversification
What diversification did in the backtest
The seven strategies finished positive as a portfolio in all 19 complete years. The study covers 6,012 closed trades in the saved Balanced v2.1.7 backtest, from 6 November 2006 to 31 December 2025. The workbook marks 2006 as a partial year.
Scroll to read all three columns.
| Area | How much it helped | Why it helped |
|---|---|---|
| Floating equity drawdown | The largest combined drop was £15,836, about 50% smaller than the £31,454 sum of the pillars’ drawdowns at that moment. The largest percentage drop was 6.77%. | Open trades across the pillars moved differently. On days when sampled equity fell, gains in some pillars offset about 14% of the losses in the others. |
| Balance drawdown | The largest combined drop was £15,319, about 52% smaller than the £32,029 sum of the pillars’ drawdowns at that moment. The largest percentage drop was 6.47%. | The pillars went through losing stretches at different times. Gains already booked by some cushioned losses elsewhere. |
| Returns | The portfolio made £340,825 net. It earned about 85% more per unit of daily balance volatility than the best individual pillar at the recorded trade sizes. | All seven pillars contributed profit overall. At least one pillar lost money in 14 of 19 complete years, while the portfolio finished positive in all 19. |
Each pillar’s drawdown is measured from its own prior peak using its recorded trades and sizes. The comparison adds those drawdowns at the time of the portfolio’s largest cash drop. The largest percentage drops occurred at different points in the test.
Floating equity was reconstructed from hourly price snapshots where quotes were available, so losses between snapshots may be missed. The comparisons use the trades from this one backtest. They cannot show what would have happened if each pillar had traded on its own.
Where the pillars helped and where they overlapped
- Every pillar had losing years. P5 had the most profitable years, finishing positive in 17 of 19.
- At least one pillar made money in 18 of the portfolio’s 23 weakest months. P7 gained in 10 of those months and was roughly flat across them combined.
- Monthly realised P&L correlations ranged from −0.17 to +0.16 across the 21 pairs. There were still 458 entries (7.6%) where another pillar held the same instrument in the same direction.
- P1 and P2 had the most same-direction overlaps and both lost money in 17 of the weakest 23 months. The overlap counts leave out position sizes and shared currency exposure across instruments.
How MultiPillar puts diversification to work
Risk control through diversification is the foundation. These features let the seven strategies operate as a coordinated portfolio.
- Seven complementary approaches
- Trend, reversal, range and volatility concepts seek opportunities in a range of market conditions, reducing reliance on a single trading approach.
- Pillar participation
- Individual strategies can be enabled or disabled, so the portfolio can be evaluated with a chosen combination of pillars.
- Individual risk weights
- Each pillar has its own risk scale, allowing its contribution to be adjusted within the shared framework.
- Portfolio evaluation
- Backtesting helps assess how the strategies behave together, including overlapping losses and total drawdown. Seven strategies do not automatically mean seven independent sources of risk.
Diversification aims to spread risk. It cannot guarantee that strategies will offset one another or prevent losses.
← Back to Risk and Safety