Fast Trend (Crisis Alpha Overlay)

Thesis (edge)

Most crises are not instant. They unfold over weeks and months. Equities fall, then bounce, then fall further. Bonds rally. The dollar or the yen strengthens. Oil collapses or spikes. A trend system with a short lookback notices these moves early enough to flip short in equities and long in bonds while the move still has a long way to run.

That is the entire idea behind crisis alpha. You are not buying insurance in the option sense, where you pay a known premium for a known payoff. You are running a strategy whose natural shape is to be positioned in the direction of any sustained move, and sustained moves are exactly what a crisis produces.

The trade-off is that the same fast reaction that makes it useful in a crisis makes it expensive the rest of the time. It gets whipsawed constantly in normal markets, and it will show a string of small negative years while everything is fine. You should expect that and budget for it rather than being surprised by it.

Where it works (regimes)

It works in drawn-out declines: 2000 to 2002, 2008, the second half of 2011, the fourth quarter of 2018, February to March 2020 and most of 2022. In each of those, the move was persistent enough for a fast signal to get positioned and stay positioned.

It does not work in a one-day crash. If the market falls 20 percent on a Monday with no warning, a trend system was long on Friday and it loses with everything else. October 1987 is the classic example. Anyone selling fast trend as protection against a sudden crash is selling something it cannot deliver.

It also does badly in a sharp V-shaped recovery, because it will have turned short near the bottom and then get run over on the way back up. The pain from a V-shaped bounce often erases a big chunk of what the strategy earned on the way down.

Signals

• Trend lookbacks in the range of roughly 10 to 40 days, which is much faster than a standard managed futures programme.
• Blend two or three speeds rather than using one, to reduce the chance that a specific number was a historical accident.
• No slow overlay. The whole point is responsiveness, and adding a 12-month filter defeats the purpose.
• Optionally include a volatility expansion trigger, so the sleeve grows when volatility is rising from a low base. This adds responsiveness at exactly the moment you want it, but it also adds another parameter to overfit.

Portfolio construction

Treat this as an overlay on a portfolio, not as a standalone fund. The natural size is a modest sleeve, perhaps 5 to 15 percent of the portfolio, run at a meaningful internal leverage so that it can move the needle when it works.

Restrict the universe to the most liquid contracts. Equity index, government bonds, major currencies, gold and crude are enough. In a genuine crisis you must be able to get in and out, and the second tier of futures markets becomes untradeable at exactly the wrong moment.

Bonds deserve careful thought. Fast trend in bonds is a large part of the crisis payoff historically, because bonds rallied in most equity crises. That relationship broke in 2022, when stocks and bonds fell together. A crisis hedge that relies on the old stock-bond relationship is not a hedge, it is a bet on that relationship persisting. The trend system handles this automatically, because it will go short bonds if bonds are falling, which is exactly what saved trend followers in 2022.

Risk model

The main risk is behavioural, not statistical. The sleeve will lose money for several years in a row. The pressure to cut it will be enormous, and it will peak right before it would have paid. Write the governance rule down before you start: this sleeve is not evaluated on standalone return, it is evaluated on what it does in the worst quartile of months for the main portfolio.

The second risk is that costs quietly eat the strategy. Fast trend has low gross returns per trade and high turnover. If your cost assumptions are 30 percent too optimistic, the entire edge disappears and you are left holding a hedge that costs more than it protects.

Costs & implementation

This is the most cost-sensitive strategy in the trend family. Turnover can be several times higher than a slow programme, and the fills come during volatile periods. Use conservative slippage and test what happens if costs double.

Execution automation matters. In a fast market the difference between a rule that says "trade at the close" and one that says "trade whenever the signal flips" is significant. Keep the rule simple, keep it scheduled, and do not improvise.

Capacity is a real constraint. Fast trend does not scale to enormous size in the markets that matter for crisis protection, which is one reason the large managed futures firms run slower systems.

Failure modes

• One-day crashes with no preceding trend, where the sleeve is on the wrong side.
• V-shaped recoveries that punish the short position established near the low.
• Death by transaction cost in normal markets.
• Getting cut by the investment committee after three flat years.
• Assuming the bond leg will always rally in a crisis.

Our Notes & Suggestions

Judge this sleeve by its conditional performance. Sort every month by the return of the portfolio it is hedging, look at the worst 10 percent of months, and ask what the sleeve did in those. If the answer is positive on average and large enough to matter, it is doing its job even if its standalone Sharpe ratio is close to zero.

Compare the annual cost of the sleeve against the annual cost of buying equity put options with a comparable payoff profile. Very often, fast trend is a cheaper way to buy roughly similar protection against slow crises, and a much worse way to buy protection against sudden ones. Owning a little of each is not a bad answer.

Do not expect it to work in every crisis. Expect it to work in most of the long ones, and be clear internally about which kind of disaster you are actually insured against.