Read benchmark, costs and drawdown together
Return without timing, a benchmark and a cost record is difficult to interpret. A useful comparison first establishes what both portfolios were able to earn under the same assumptions.
Trace the first execution
In the synthetic four-day example, ETF_A rises 100→110→121→108.9 while ETF_B stays at 100. Both portfolios start at 10,000. The signal is formed after day one’s close and executed at day two’s close. The 100→110 rise is therefore excluded.
| Checkpoint | 60/40 strategy | 50/50 buy-and-hold benchmark |
|---|---|---|
| Initial value | 10,000.00 | 10,000.00 |
| Day two, after assumed initial cost | 9,990.00 | 9,990.00 |
| Day three | 10,589.40 | 10,489.50 |
| Day four | 9,930.06 | 9,940.05 |
| Cumulative return | -0.70% | -0.60% |
| Maximum drawdown | -6.23% | -5.24% |
Assumed friction is 10 basis points per side. Initial turnover is one for both; later turnover uses absolute changes in risk-asset weights, without counting cash again. The benchmark pays the same initial cost and then does not rebalance.
Recompute the risk measure
Return is final value divided by initial value minus one. Drawdown at each observation is current value divided by its running peak minus one, including the initial value and post-cost observations. Strategy drawdown on the last day is 9,930.06/10,589.40−1, approximately -6.23%.
A strategy can finish near its starting value after a larger interim loss. That is why drawdown is not replaced by cumulative return. Four observations cannot establish reliable long-run risk or justify annualized marketing.
Find a discrepancy before explaining it
If a report differs, compare the first divergent daily row. Check initial execution, cost, weights and index ratios before discussing investment logic. Comparing strategies with different warm-up periods requires a separate common interval; two full-period cumulative returns are not automatically comparable.
What the result means
This constructed strategy loses slightly more than its benchmark. The result does not become a failure because it is unattractive. It tests the stated model. Investor.gov’s ETF overview provides vehicle context; these hypothetical results are not sourced fund performance. Real fills, taxes, liquidity and lot sizes remain outside the weight model.
FAQ
Does a lower return mean the tool failed?
No. Failure means wrong computation or unmet inputs, not an unattractive valid result.
Can I compare annualized returns from this sample?
No. The four observations are a timing test, not long-term evidence.