In Rob Carver’s Advanced Futures Trading Strategies (Amazon,) there’s a chapter, “Strategy Nine: Multiple Trend Following Rules,” that uses composite trend-following rules to drive a long-short strategy. We explored the strategy through an Indian market participant’s lens.
We ran Strategy 9 across five increasingly broad universes — NIFTY indices, a 15-instrument multi-asset basket, 21 crypto coins, a dynamic walk-forward crypto universe, and MSCI country equity indices.
Scaled long-only is the only variant worth keeping
Across every experiment, scaled long-only (position size proportional to forecast strength) produced the best risk-adjusted returns. The short side of binary long-short strategies consistently loses money or adds drawdown without compensating Sharpe improvement. Binary long-only earns higher raw returns in some universes but with drawdowns that rule out leverage — making it strictly inferior to buy & hold on an absolute return basis.
As a retailer, scaling long-only cash positions is feasible if you automate everything. Strategy 9 is practical in that sense.
The strategy output is unleverageable
This is the central failure. The whole point of trend-following is to produce a return stream smooth enough that you can apply leverage and beat buy & hold. Strategy 9 never achieves this. Even its best variants have drawdowns in the 25–60% range. At 2× leverage, a 30% unlevered drawdown becomes 60% — a portfolio killer. Without the ability to lever up safely, you trail B&H on absolute returns.
Indians are anyway prohibited by regulations to take on leverage in international markets. However, there were only a couple of Indian indices with futures where leverage on Strategy 9 was workable. This is probably the most disappointing result of our backtests.
Expanding the universe yields diminishing returns
The Big 3 crypto coins (BTC, ETH, SOL) already capture most of what trend-following can extract from crypto. Adding 18 more coins, or dynamically re-selecting from the full universe each month, adds complexity without improving the portfolio. The best-performing 3–5 instruments drive the results; the rest contribute noise or outright negative returns.
Trading crypto for as an Indian is anyway not feasible given the current tax structure. Not sure if we lose out on much here.
Performance may be period-dependent
The MSCI equity experiment revealed that yearly returns effectively stopped working around 2009. What looks like a decent full-sample Sharpe may be entirely back-loaded — an artifact of pre-2009 returns that never recurred. The same question hangs over the other universes; we did not slice them the same way.
Cost screens hurt; inverse-vol weighting doesn’t help
Carver’s cost screen eliminates the faster filters (EWMAC2, EWMAC4), making the strategy sluggish and cutting returns 30–40% without meaningfully improving risk metrics. Inverse-volatility weighting, which should theoretically down-weight volatile losers and improve risk-adjusted returns, made no material difference versus simple equal-weight.
The best risk-adjusted result is narrow, scaled, long-only
A 50-50 equal-weight blend of Scaled Long-Only on NIFTY MIDCAP 150 and NIFTY SMALLCAP 250 produced a Sharpe of 1.18 with a 13.1% drawdown — the cleanest equity curve across all experiments. But even this trails B&H on unlevered absolute returns. At 2× leverage it earns 18.9% annualized with a 26% drawdown — the closest we got to a genuinely usable strategy.
Bottom line
Strategy 9 does not survive real-world scrutiny. It finds trends, it earns positive Sharpe, but it cannot produce a return stream smooth enough to lever into genuine outperformance. The drawdowns are always too deep, the universe expansion never helps, and the equity index variant suggests whatever edge existed may have expired. This is a strategy that looks promising in theory and in plots — and fails on closer inspection.
Code and charts on github.
