Benchmarking methodology
How OLTA evaluates baskets against published references. Written so it works whether or not you have met a Sharpe ratio before: the primer comes first, the two references second, and the reading key last.
What these numbers measure
Three numbers carry this page. Two of them describe a basket on its own, the third describes the worst moment inside the window. Everything after this section is a comparison of those three between a basket and a reference.
- Sharpe ratio
- Return earned per unit of total volatility · the standard risk-adjusted yardstick. Higher is better. Above 1 is strong, above 2 is excellent. It uses total volatility, so it penalises big upside swings too.
- Annualised volatility
- How much the basket's returns swing, scaled to a yearly figure. Lower is steadier. It's the denominator of the Sharpe ratio.
- Max drawdown
- The deepest peak-to-trough fall over the backtest window. Closer to 0 is better · it's the worst loss someone who bought at a peak would have ridden through before recovery.
(R − r_f) / σσ_annual = σ_daily · √252- Below 0Lost money over the window. The volatility bought nothing.
- 0 to 0.5Positive, but thin. The return is small next to the swings it took.
- 0.5 to 1Respectable. Roughly what a decent long-only book prints.
- 1 to 2Strong. Return is compounding faster than risk is being taken.
- Above 2Excellent, and rare. Worth checking the window length before believing it.
A Sharpe measured over a short window is noise rather than a weak signal, so OLTA withholds the grade entirely below 90 days of history and prints "Not graded" in its place. Every figure on this surface is a backtest on simulated funds.
Why two benchmarks
No single benchmark answers both questions an allocator asks of an actively-constructed basket, so OLTA publishes two.
Did you beat just holding Bitcoin?
- Reads
- The dominant crypto risk premium.
- Holdings
- 1 asset
- Rebalance
- None. It is one asset.
Did your construction earn its keep against naive diversification?
- Reads
- The equal-weight 100 baseline.
- Holdings
- top 100 liquid crypto assets
- Rebalance
- Monthly reconstitution.
An active basket competes against both. This mirrors the convention used by institutional crypto research desks, where the BTC line and a broad-basket line are reported side by side on every fund factsheet.
BTC reference
Bitcoin is the institutional default for crypto performance comparison. Its risk-adjusted profile depends materially on the window an analyst selects: Sharpe over inception is not Sharpe over the last five years, and the gap between the two is wide enough to change a conclusion. All four canonical anchors are published below so the comparison is honest about start-date convention. The headline figure on each Backtest panel uses the rolling 5y window, the convention adopted by Galaxy Digital research and the Bitwise BITW factsheet.
Sharpe by anchor
Bitcoin's risk-adjusted profile depends materially on the window chosen. The figures below are computed from the same daily OHLC the OLTA backtest pipeline consumes, with zero risk-free rate and 252 trading-day annualisation.
| Anchor | Window | Years | Sharpe | Ann return | Vol | MaxDD |
|---|---|---|---|---|---|---|
| 2010-07-17 | Inception | 15.9y | 1.68 | 145.7% | 86.6% | −93.1% |
| 2014-01-01 | Post-mania | 12.4y | 0.59 | · | · | · |
| 2017-08-17 | Binance era | 8.8y | 0.69 | 39.1% | 56.8% | −83.2% |
| 2020-01-01 | Rolling 5y | 6.4y | 0.87 | · | · | · |
OLTA Crypto Benchmark · Equal-weight 100
OCBE100 is an equal-weighted composite of 100 liquid crypto assets from the OLTA universe, each held at one percent. Stablecoins, gold-pegged tokens and the lowest-liquidity long tail are excluded so the basket stays tradeable at institutional scale. Reconstitution is monthly: the constituent list freezes for the month and is reviewed on the first business day. Its level is computed by the same divisor method as any other basket, so the benchmark and the baskets it measures share identical arithmetic.
OCBE100 is a control group, not a target. It tells an allocator what an undifferentiated long exposure to the broad market would have returned over the same window. An active basket that does not clear OCBE100 net of rebalance cost is not earning its construction premium: either the weighting scheme adds no risk-adjusted return, or the constituent universe is the wrong one for the regime. Either finding is useful, which is why the reference is published rather than quietly dropped when it flatters less.
Window alignment
Every vs-BTC and vs-OCBE100 delta is sliced to the basket's own backtest window, never to a global fixed window. A basket with 67 days of history is compared against the same 67 days of BTC and OCBE100. A basket with five years of history is compared against the same five years.
Mixing windows would contaminate the delta with a regime difference that has nothing to do with the construction being evaluated. A 2020 anchor captures a different macro regime than a 2024 anchor, and the resulting Sharpe delta would mostly be reading that mismatch rather than any property of the basket.
The same rule applies across timeframe tabs. When the Backtest panel switches to the 6m or 3m view, the BTC and OCBE100 references shown alongside are recomputed on that same trailing window. A basket can look like a hero against BTC on the 2y window and a laggard on the 3m window, not because the construction changed but because the benchmark moved differently in those two regimes. Reading both is the point.
How to read the deltas
The vs-benchmark strip reports three numbers per reference: total-return delta, Sharpe delta, max-drawdown delta. The first two read in the obvious direction, higher is better. The third reads the other way, and that inversion is where most misreadings happen: a drawdown carries a minus sign, so a delta that is more negative means the basket fell further than the reference did.
Return and risk have to be read together. The four combinations below cover every strip you will meet.
- Sharpe higherDrawdown shallowerImproved risk profile
Better return per unit of risk and a shallower fall than the reference. The combination an allocator is looking for, and the one that survives a change of window most often.
- Sharpe higherDrawdown deeperPaid for with depth
More efficient overall, but it fell further at the worst moment. Acceptable when the Sharpe gap is clear and the holding horizon is long enough to sit through the pit.
- Sharpe lowerDrawdown shallowerDefensive, not additive
Held up better in the fall but earned less per unit of risk. Useful as ballast next to a BTC-heavy book, weak as a standalone allocation.
- Sharpe lowerDrawdown deeperNot earning its construction
Worse on both counts. The weighting scheme is adding risk without adding return over this window, and the reference is the cheaper way to hold the same exposure.
Read the two references for different things. BTC tells you about the crypto beta you bypassed. OCBE100 tells you whether the diversification you paid for showed up.