Market Regime: Every Strategy Is Built for One Kind of Weather

A racing dinghy is a superb boat. On a bright afternoon with a steady breeze it will outrun almost anything on the water. Take it out in a gale and it comes back in pieces, if it comes back at all. Nobody on the quay concludes that the dinghy was badly designed. They conclude that someone sailed it in the wrong weather.

Markets have weather too, and it has a name: the market regime. Trending or sideways, calm or violent, euphoric or afraid. Samuel Rondot spent years watching traders blame the boat for the storm.

His argument, made across six articles between 2011 and 2014, is easy to state and hard to live by. A strategy is not good or bad. It is good in one kind of weather. Your job is not to find the boat that wins in every sea, because that boat does not exist. It is to know which weather your boat was built for, to read the forecast before you cast off, and to stop treating a storm as proof that the hull is broken.

You are betting on behaviour, not on price

In August 2014, Samuel published a short piece built around one sentence. He had spent ten years trying to put it into words, he wrote, and a writer he names only as Charles handed it to him in an article of his own:

"You're not betting on price levels. You're betting on the persistence of price behaviour. That's not the same story."

Read it twice, because it moves the whole question. A strategy has no idea where the price will be tomorrow. What it has is evidence that when a certain behaviour shows up, a trend that keeps running, a range that keeps bouncing, a burst of volatility that keeps expanding, doing the same thing again and again has paid more than it cost. Samuel's own definition, in the same piece: a trading strategy is a mathematical edge, proven on a given market, that makes money by doing the same thing over and over in the same context.

If the edge lives in the behaviour, the behaviour has to be present for the edge to pay. Which leaves the question every losing streak eventually forces on you: what happens when it is not?

No boat sails every sea

Two days later, Samuel answered it. When the market is falling or rising, moving a lot or barely at all, when Wall Street is open alongside Europe or is not, when it has just spent a month falling or a month climbing, he wrote, the market IS DIFFERENT. Trading it the same way in all of those conditions is not a style. It is a contradiction.

You can name the main weather systems. In a trending market, price keeps moving in one direction, and a trend-following strategy is the boat built for it. In a sideways market, also called a ranging market, price swings back and forth inside a band, and counter-trend strategies built on mean reversion are at home. The weather that sinks the most traders is the third: a choppy market that flips between the two too often for either boat to settle. The trend strategy is whipsawed at every turn. The counter-trend strategy is caught each time a bounce turns into a run.

So what of the boat that claims to sail all three? Samuel's answer is that you have seen it many times, in the equity curves that never falter on the sites that sell them. It is easy to build if you only look backwards: bend a method with enough mathematics and it never lost in the past. Trade it live and you discover that you painted a perfect curve onto history, one that predicts nothing. His alternative runs the other way. Build each strategy as simply as possible, aimed at one behaviour, and accept that it will stack up false signals whenever that behaviour is absent.

That leaves you with a boat that is fast in one weather and slow in every other. The next question is how you tell which weather you are in before you leave the harbour.

Read the weather before you read the signal

Samuel's clearest lesson on this came from the economic calendar. Take the monthly US jobs report, the non-farm payrolls figure published on the first Friday of most months. What moves the market, he argued in April 2013, is never the number itself but the gap between the number and what analysts expected, and even that gap is filtered through the weather it lands in.

In 1999, with the economy running hot, a jobs figure that came in too strong would knock the market down, because strong hiring meant rising wages and rising wages meant inflation. In April 2013, an average figure dropped into a market that had been euphoric for weeks cost it barely one percentage point, for a few hours. The rule he drew from it: the two biggest reactions always come from bad news in a pessimistic market and good news in an optimistic one. The other pairings carry little weight.

Then he turned the lens on the hundreds of client accounts he watched. The most basic adaptation to the weather, working only the upside in a euphoric market and only the downside in a crisis, was one almost none of his clients ever showed. Yet step back from the screen for a moment and you know perfectly well what weather you are in. The failure is not perception. It is that the forecast never makes it into the trade.

How does the forecast make it into the trade? Through a filter, kept apart from the trigger.

Filter first, then pull the trigger

In June 2014, Samuel set out the method he had come to rely on most: your strategy has to start with a FILTER. The filter defines the type of market you are in and puts a number on it. Only then do you decide what to do inside it.

The filters he used were deliberately plain. The direction of the trend. The level of volatility. The time of day: European mornings, he observed, tended to correct the excesses of the previous US close, which suited counter-trend trading, while afternoons, when Europe follows the US, suited trading with the trend. Inside each filter sits a separate trigger. If volatility is low, buy on a small acceleration. If it is high, wait for a sharp one.

The separation is the point. A moving average crossing is an entry like any other, but it tells you nothing about the weather it happened in. The filter is the forecast. The trigger is the moment you slip the mooring. Merge the two and you end up sailing because the clock said so, not because the wind did.

Keeping them apart pays twice. Signals become easier to find, because you only look for them in conditions you have already defined. And diversification becomes deliberate, because a counter-trend boat and a trend boat can share a portfolio, each one sailing its own weather. Every one of those boats still has a running cost, which what each trading style costs you lays out style by style.

The case against filtering

The strongest objection to all of this was written by Samuel himself, less than three months later. If a strategy cannot trade properly in certain conditions, why not filter those conditions out, so that it only trades when the weather suits it? Because, he answered in September 2014, a filter is late by nature. For it to tell you the market is rising, the market has to have risen already. For it to spot the end of a rally, the rally has to have ended. A filter that switches a strategy on late and off late misses the first good trades and holds the last bad ones, and as a rule it does more damage than the problem it was built to prevent.

This holds, and it does not contradict June. The two filters do different jobs. The June filter chooses which kind of trade fits the weather you can already see. The September filter tries to keep a single boat in harbour whenever its weather turns, and you only see the turn after the rain has started. Choosing the boat for the forecast works. Docking it every time a cloud passes costs you the first hours of every sunny spell and the last hours of every storm.

Which means that for much of its life, your strategy will be at sea in weather it was not built for. How do you live with that without abandoning it at the worst moment?

A storm is not proof the hull is broken

He learned how long a storm can last the hard way. In January 2011, looking back on a year in which his own trend systems on the CAC 40 found almost nothing to hold, he wrote that the best trend-following system in the world still needs a trend, and that a market can stay rotten for a year, perhaps for two or three.

His answer is patience, and he was unambiguous about why. Every strategy, sooner or later, runs into stretches of market it is simply not suited to. It loses then not because it is bad, but because the market's behaviour has left its comfort zone. The behaviour comes back round. Switch the strategy off in the storm and you are relying on yourself to switch it back on in time for the sun, which is the same late filter in human form.

Patience is not a licence to ignore a broken boat, though, and the difference can be tested. Samuel never wrote the test down as such, but it follows from everything above. Ask whether the weather your strategy was built for is present. If a trend strategy is losing in a sideways market, that is the storm, and it will pass. If it is losing while the market trends cleanly, the weather is right and the boat is not, and no amount of waiting will fix it.

His other safeguard was structural. A good filter has to be simple, and it has to be tested over a period long enough to trust the result, because a filter with dozens of conditions describes the past precisely and the future badly. That is why he preferred several simple systems to one complicated one: less risk of fitting the past, and a fleet of different boats that smooths the voyage when any one of them is becalmed. His last word on the subject, in 2013, was addressed to traders who never analyse their own results: THERE IS NO OTHER WAY TO SUCCEED. Learn. Correct. Improve.

The weather does not stop when the market never closes

Samuel's time-of-day filter rested on a market that opened and closed, with Wall Street joining Europe each afternoon. A perpetual futures market has no closing bell. It is tempting to conclude that a market which never closes has only one weather. It does not. Trends still form and fade, ranges still hold and break, volatility still expands and contracts. The market regime still changes. What disappears is the night: the pause that used to end every session whether you were ready or not.

That makes the forecast more important, not less. If you run strategies through algorithmic trading, the regime question stops being a feeling and becomes a line of logic you have to write on purpose: the filter that chooses the boat, kept apart from the signal that casts off. Samuel's two warnings travel with it. Keep the filter simple enough to survive a future that will not repeat the past. And do not use it to drag the boat into harbour every time a cloud crosses the sun.

Market regimes: questions answered

What is a market regime?

A period in which the market behaves in a consistent way: trending, moving sideways in a range, or swinging choppily between the two, at high or low volatility. A strategy is usually built to profit from one regime and gives back part of that profit in the others.

How do you identify the current market regime?

With a simple filter defined before you look for a trade: the direction of the trend, the level of volatility, or the time of day. Samuel's rule was that the filter must stay simple, because a filter with many conditions describes the past precisely and the future badly.

Should a strategy stop trading when the regime changes?

Rarely by being switched on and off. A filter that detects a regime change is late by construction, so it tends to miss the first trades of the good weather and hold the last trades of the bad. A sturdier answer is to run several simple strategies, each built for a different regime, side by side.

What is the difference between a trending, a sideways and a choppy market?

A trending market keeps moving in one direction, which rewards trend-following. A sideways market swings inside a range, which rewards counter-trend trading. A choppy market flips between the two too often for either approach to settle, and both can lose in it at the same time.

Read the series

Each article below was published by Samuel Rondot between 2011 and 2014. The originals are linked at the top of each page.

• What I Learned in 2010 (8 January 2011)
• The Quality of Statistics (10 April 2013)
• How to Give Yourself a Better Chance of Success When Opening a Position (18 June 2014)
• The Sentence That's Going to Transform Your Trading (27 August 2014)
• You Always Have to Trade the Same Way: True or False? (29 August 2014)
• The Value Added by a "Filtered" Strategy (3 September 2014)

Trading leveraged products carries a high risk of loss. The accounts above are one trader's observations between 2011 and 2014, not studies and not forecasts.