The RSI-2 Strategy: Settings and Rules for Mean Reversion

The RSI 2 strategy is a short-term mean reversion system built by Larry Connors, and it does the opposite of almost everything new traders are taught about RSI. Instead of the default 14-period setting, it cranks the lookback all the way down to 2. Instead of selling when RSI hits 70, it buys when a stock that is already in an uptrend gets briefly slammed to an extreme low. The whole idea is simple: in stocks and stock indices, sharp short-term drops inside a larger uptrend tend to bounce back, and a 2-period RSI is sensitive enough to flag those drops the moment they happen.

This article lays out the exact settings and rules so you can test the RSI 2 strategy yourself, plus the parts most write-ups skip: why the trend filter is the whole game, why Connors argued against stops, and where this setup quietly breaks. None of it is magic, but the rules are precise, and precision is the point.

What the RSI 2 strategy actually is

Connors introduced the approach in his book "Short Term Trading Strategies That Work" after running tests across hundreds of thousands of trades on stocks and indices. The core finding was that a 2-period RSI does a far better job of timing short-term entries than the standard 14-period version most platforms ship with. A 2-period RSI is extremely jumpy. It can swing from 90 to 5 in a couple of bars. That jumpiness is a feature here, not a bug, because you are trying to catch fleeting moments of panic, not smooth out the noise.

One thing to be clear about up front: this is not a counter-trend strategy that tries to call tops and bottoms. It is a trend-following strategy that uses a mean reversion entry. You only buy dips inside an existing uptrend, and you only short rips inside an existing downtrend. The 200-day moving average decides which side of that line you are on, and you never fight it.

The exact settings for the RSI 2 strategy

Everything runs on daily bars. There are only three indicators on the chart, and two of them are just moving averages.

That is the entire toolkit. A 2-period RSI, a 200-day SMA, and a 5-day SMA. You do not need volume, you do not need a second oscillator, and you do not need to draw a single trendline. If you want to scan a watchlist for names that just printed an RSI(2) reading below 5 or 10 above their 200-day line, that is exactly the kind of mechanical filter ChartRead can run for you so you are not eyeballing fifty charts by hand.

The long entry rules, step by step

Here is the full long side, in order. All three conditions have to be true on the same day before you act.

1. Price is trading above its 200-day simple moving average. This confirms the larger trend is up. No exceptions, no "it is only a little below."
2. The 2-period RSI closes below 10. This is the short-term oversold trigger. Connors tested every threshold from 0 to 10 and found that the lower you set it, the better the average return per trade, just with fewer signals. Below 5 is the more aggressive, higher-conviction version.
3. You enter on the close of that day, or the next open.

To exit, you wait for price to close above its 5-day simple moving average. That is the entire exit. When the short-term average is reclaimed, the bounce you were betting on has happened, so you take the trade off. Holds are typically short, often just a few days.

The short side is a mirror image

The short setup flips every condition. You are now looking for a brief pop inside a downtrend.

1. Price is trading below its 200-day simple moving average, confirming a downtrend.
2. The 2-period RSI closes above 90 (or above 95 for the stricter version).
3. Enter short, then cover when price closes below its 5-day simple moving average.

In practice most traders run the long side only. Indices and large-cap stocks spend far more time above their 200-day average than below it, so long signals are more frequent and the upward drift of the market is working with you instead of against you. The short side exists and tests fine, but it is the harder half to trade live.

Scaling in: the part people forget

The version of the RSI 2 strategy that tests best is not a single all-in entry. Connors used a scaling approach. The deeper the oversold reading goes, the more you add. A common structure looks like this: buy a first unit when RSI(2) drops below 10, then add a second unit if it pushes below 5 on a later bar, and so on. You are averaging into weakness on purpose, because in a confirmed uptrend, a deeper short-term flush historically meant a stronger snap-back.

Why Connors said no stops, and why that needs a caveat

This is the most controversial rule. In his testing across stocks and stock indices, Connors found that adding a fixed stop loss actually hurt total returns. The reason is structural: this strategy buys things that are already falling fast. A tight stop sits right in the path of the very volatility you are trying to exploit, so you get stopped out at the local low again and again, right before the bounce you were waiting for.

That said, "no stops" is not the same as "no risk control." A mean reversion system with no stop assumes the position will eventually revert, and once in a while it does not. A single name can gap down on earnings or news and keep going. So the real risk management here lives in other places: the 200-day filter that keeps you on the right side of the trend, small position sizing per trade, trading liquid instruments rather than thin small-caps, and capping how many of these correlated dip-buys you hold at once. If you are not comfortable trading without a hard stop, use a wide disaster stop well below your average entry rather than a tight one, and accept that it will slightly lower the backtested edge.

One filter that cuts false signals

An optional refinement: instead of buying the instant RSI(2) prints below your threshold, wait for it to cross back up through 50 before entering. This makes you wait for the first sign that the falling knife has actually turned, at the cost of a slightly worse entry price. It will not change the strategy's character, but it does trim some of the trades where price kept sliding after the first oversold reading. Test it both ways on your own data and see which fits your tolerance for drawdown.

Where the RSI 2 strategy breaks

No edge works everywhere, and it is worth knowing the failure modes before you risk money.

• Strong sustained downtrends. Mean reversion assumes price comes back. In a 2008 or 2022 style grind lower, oversold gets more oversold for weeks. The 200-day filter is what mostly keeps you out, but a fast crash can drop price below the 200-day right as you are holding.
• Individual stocks with real news. The strategy was built and tested largely on indices and broad baskets, where one company's blowup gets diluted. A single stock that is oversold for a reason, like a guidance cut, may not revert at all.
• It is crowded. This is a 15-plus-year-old published strategy. The raw edge on major indices is thinner than it was in the book's backtests. It still works as a framework, but do not expect the eye-popping equity curves from a decade ago.

How to put it into practice

Start by backtesting it on a liquid index ETF or a basket of large-caps before you ever go live, and run the numbers yourself rather than trusting anyone's printed results. Pick your aggressiveness, below 10 for more trades or below 5 for stronger ones, decide long-only or both sides, and define exactly how you will size and scale. Then paper trade it for a few weeks so you feel what it is like to buy something that looks terrible on the day you buy it, because that discomfort is the whole reason the edge exists. The rules are mechanical for a reason: the moment you start overriding them based on how a chart "feels," you are no longer running the RSI 2 strategy, you are just guessing with extra steps.