Mean reversion is the mirror image of trend followingTrend FollowingAn approach that buys what is rising and sells what is falling, riding moves until they reverse.Click the word to learn more. When an instrument has run unusually far in one direction, mean reverters bet that it will swing back toward its average. Sell the rip, buy the dip, take the profitProfitThe gain or loss on a closed position, once spreads, commissions, and swaps are taken out.Click the word to learn more when the move retraces.
The shape inverts everything. Mean reversion strategies typically win more than 60 percent of their positions. Most of those wins are small: the move retraces, the algorithm closes out, the next setup arrives. The catch is the lossLossThe realized loss on a closed position, or the sum of every losing position across a period.Click the word to learn more profile. When a real trend starts and refuses to revert, a mean-reverter takes a single large loss that can erase weeks of accumulated small gainsGainsThe sum of every winning position in a period, before losing positions are subtracted.Click the word to learn more.
This is the trap of looking at mean-reversion strategies on a calm equityEquityThe live value of your broker account, including the floating profit or loss of open positions.Click the word to learn more curve. Six clean months of small green wins are easy to find. The seventh month, when a trend develops and the algorithm gets caught fading it, is the one that defines the long-run profile. The strategy that has not yet had a bad period is not a strategy without bad periods.
Good mean-reversion algorithms ship with regime filters: volatility, time of day, correlation breakdown, anything that signals "this is no longer a ranging market." The filters do not eliminate the risk. They prune the worst conditions. Javlot strategies of this style expose the filters and their historical behavior per regime.