RSI 2 Strategy: Larry Connors’ Short-Term Pullback Rule, Worked Through

Most people meet RSI as a 14-day line that wanders between 30 and 70. Then someone shows them the RSI 2 strategy and the same indicator suddenly looks like a different animal. Shrink the lookback from 14 days to two and RSI stops drifting. It swings from single digits to the high 90s in a matter of sessions. Larry Connors built a short-term pullback rule around exactly that jumpiness, and it’s one of the cleaner mean-reversion setups you can study.

This piece walks through the four rules as StockCharts’ ChartSchool describes them, then runs the maths on a hypothetical stock so you can see how a signal forms, where the trade gets out, and what the popular “wait for 50” filter costs you. The numbers in the example are made up for teaching. The rules are not.

What the RSI 2 strategy actually is

The ChartSchool write-up on RSI(2) calls it a fairly simple mean-reversion strategy, developed by Larry Connors, that buys or sells after a corrective period. The key phrase is “within an ongoing trend”. This is not a bottom-fishing tool. ChartSchool says plainly that it isn’t designed to identify major tops or bottoms. The idea is to buy a dip in a stock that is already going up, and short a bounce in one that is already going down.

Connors’ own one-liner, as ChartSchool summarises it, is that traders should buy pullbacks, not breakouts, and sell oversold bounces, not support breaks. If you’ve ever bought a breakout at the top of a candle and watched it fade the next morning, you’ll understand why that idea appeals.

If RSI itself is new to you, start with our guide to reading the RSI indicator. Here’s the short version.

A quick refresher on RSI

The Relative Strength Index was developed by J. Welles Wilder Jr. and introduced in his 1978 book New Concepts in Technical Trading Systems, according to Investopedia’s RSI explainer. It compares average gains with average losses over a lookback period and squeezes the result into a 0 to 100 scale:

RSI = 100 − 100 / (1 + average gain ÷ average loss)

Investopedia notes that 14 periods is the standard setting, and that the classic reading treats 70 or above as overbought and 30 or below as oversold. With 14 days in the average, one bad session barely dents the number. With two days, one bad session can drag it into single digits. That sensitivity is the whole point of RSI(2).

So the thresholds move too. ChartSchool describes RSI(2) below 10 as deeply oversold and above 90 as the zone for short-selling ideas.

The four rules, one at a time

1. Let the 200-day average decide the direction

Connors uses the 200-day simple moving average as the trend filter. Price above it means the long-term trend is up, so you only look for buys. Price below it means you only look for shorts. This single rule kills a lot of bad trades. An RSI(2) reading of 3 in a stock that’s been sliding for six months below its 200-day line is not a bargain. It’s a falling knife with a low number attached.

2. Wait for an extreme reading

Connors tested buy levels between 0 and 10 and sell levels between 90 and 100. Per ChartSchool, returns on longs were higher when buying a dip below 5 than below 10, and returns on shorts were higher when selling a surge above 95 than above 90. The deeper the short-term washout, the better the bounce tended to be in his tests. That’s why you’ll see traders quote 5 and 95 as the working levels.

3. Decide when to place the order

You can enter just before the close or on the next open. Connors prefers the close. ChartSchool points out the trade-off: buying before the close leaves you exposed to the next morning’s gap, which can help or hurt, while waiting for the open gives you a chance at a better price. For Indian traders, this matters more than it sounds. Plenty of news, from overnight moves in global markets to results announced after hours, can land between our close and the next open.

4. Exit on a close above the 5-day average

In Connors’ S&P 500 example, longs are exited on a move above the 5-day SMA and shorts on a move below it. That makes this a genuinely short-term system. Trades often last only a few sessions.

RSI(2) is not a bottom-fishing tool. It buys a dip in a stock that is already going up, and nothing else.

The part most people skip: no stop-loss

Here’s the bit that surprises everyone. ChartSchool reports that Connors does not advocate stops. In his quantitative testing, which ChartSchool says involved hundreds of thousands of trades, stops actually hurt performance on stocks and stock indices.

We’d be careful with that. ChartSchool itself adds the caveat that trading without stops can produce outsized losses and large drawdowns, and that it would be prudent to build an exit and stop-loss plan for any system. A backtest on a broad US index is one thing. A single Indian mid-cap that gets hit by a governance story and opens 15% lower is another. Testing that says stops “hurt on average” doesn’t protect you from the one trade that wipes out ten winners.

A sensible middle path is a wide disaster stop: far enough away that ordinary noise won’t trigger it, close enough that a genuine collapse can’t sink the account. Our risk management guide covers how to size positions around a stop, and we use that method in the example below.

Worked example: a hypothetical RSI(2) long trade

Let’s take an imaginary large-cap, call it Stock X, trading at around ₹1,520. Assume its 200-day SMA is at ₹1,410, so price is comfortably above it and only long signals count. These are hypothetical closing prices for ten sessions, with RSI(2) worked out using Wilder’s smoothing.

How the smoothing works with a two-period lookback: the first average gain and average loss are the simple averages of the first two daily changes. After that, each new average is (previous average × 1 + today’s gain or loss) ÷ 2. So every day the latest move carries half the weight. That’s why the line jumps around so much.

Walk through it session by session.

• Days 1 to 3: the stock edges up from ₹1,520 to ₹1,531. Two up days and no losses means average loss is zero, so RSI(2) is pinned at 100.
• Day 4: a ₹14 drop. One down day is enough to pull RSI(2) all the way to about 28. A 14-day RSI would barely notice.
• Day 5: another fall, to ₹1,504. RSI(2) is about 12.1. Oversold by normal standards, but not below 10, and well short of the 5 that Connors’ tests favoured. No trade yet.
• Day 6: a sharper ₹18 drop to ₹1,486. RSI(2) drops to about 4.7. Price is still above the 200-day SMA. That’s the signal. Following Connors’ preference, we buy near the close at roughly ₹1,486.
• Day 7: a bounce to ₹1,499. The 5-day SMA is ₹1,507.4 and the close is below it, so we hold.
• Day 8: the stock closes at ₹1,512. The 5-day SMA is now ₹1,503.6. Close above the average means exit.

Two sessions in the trade, ₹26 a share, roughly a 1.75% move from entry. That’s what a typical RSI(2) trade looks like when it works: small, quick, and dependent on a lot of them adding up.

Sizing it with a disaster stop

Say the account is ₹5,00,000 and we’re willing to risk 1% of it, or ₹5,000, on any single trade. We put the disaster stop at ₹1,440, about 3.1% below the ₹1,486 entry. That’s ₹46 of risk per share.

• Shares = ₹5,000 ÷ ₹46 = 108.7, rounded down to 108 shares.
• Position value = 108 × ₹1,486 = ₹1,60,488.
• Actual risk if the stop is hit = 108 × ₹46 = ₹4,968.
• Gross profit at the day 8 exit = 108 × ₹26 = ₹2,808, before brokerage, STT and other charges.

Notice the shape of it. We risked close to ₹5,000 to make ₹2,808. On a single trade, that’s a poor reward-to-risk ratio. RSI(2) only makes sense if the win rate is high enough to carry it, which is exactly why Connors’ testing focused on probabilities over thousands of trades rather than any single setup. If you take this strategy seriously, you need your own records over a large sample, not a few memorable winners.

We risked close to ₹5,000 to make ₹2,808. RSI(2) only works if you win often enough to carry trades like that.

The “wait for 50” filter, and what it costs

ChartSchool flags the obvious weakness: RSI(2) signals are often early. A stock that’s fallen hard can keep falling after RSI(2) dips below 5. Its suggested tweak is to wait for some sign the move has turned. One version is to wait for RSI(2) to climb back above its 50 centreline before buying (or drop below 50 before shorting).

Look at what that does to our example. On day 7, RSI(2) is about 49.4, a whisker below 50, so the filter says wait. On day 8, RSI(2) jumps to about 73.9 and the filtered trader buys at ₹1,512. But day 8 is also the day the unfiltered trader exits, because the close is above the 5-day SMA. In this case, the filter got you in exactly where the original system got out.

That doesn’t make the filter useless. In a stock that keeps falling for four or five sessions after the first signal, it would have kept you out of most of the pain. It’s a trade-off between fewer losing entries and giving up some of the bounce. You won’t know which side of that trade-off suits you without testing on the stocks you actually trade.

Running it on Indian stocks: things to watch

• Results season. ChartSchool’s Google example shows how gaps around earnings wrecked some signals. In India, quarterly results often come out during or after market hours, so check the results calendar before buying an RSI(2) dip.
• Liquidity. A strategy that exits in two or three sessions depends on getting filled near your price. Stick to liquid names where the bid-ask spread is tight.
• Costs. With gains of a percent or two per trade, brokerage, STT, exchange charges and slippage take a real bite. Run your numbers net of costs.
• Circuit limits. Stocks in tighter price bands can lock in a circuit, and then neither your exit nor your stop gets filled.
• The short side. A short signal can need holding for several sessions, so think about how you’d carry that position before you act on one. Traders who use futures for it also take on lot sizes and margin requirements.

RSI(2) sits in the same family as other mean-reversion approaches. If the idea of buying dips inside an uptrend appeals, our piece on mean reversion trading on the Nifty covers the wider logic, and the Stochastic RSI is another way traders make RSI faster.

A checklist before you take a signal

1. Is price above the 200-day SMA (for a buy) or below it (for a short)?
2. Has RSI(2) closed below 5 (or above 95), or are you settling for 10 and 90?
3. Are results, a board meeting or another known event due in the next few sessions?
4. Where is your disaster stop, and does the position size keep the loss within your limit?
5. Do you exit on the 5-day SMA cross, a trailing stop, or an overbought RSI reading? Decide before you enter.
6. Are you logging every trade, so you can judge the strategy on dozens of results rather than one?

The short version

The RSI 2 strategy takes a 14-day indicator, cuts it to two days, and uses the resulting jumpiness to buy sharp dips in uptrends and short sharp bounces in downtrends. Trend comes from the 200-day SMA, the trigger is an RSI(2) extreme (below 5 or 10, above 90 or 95), entry is near the close or the next open, and the exit is a close back across the 5-day SMA. Connors’ tests found stops hurt performance, but a sensible trader still keeps a wide disaster stop and sizes around it. The trades are small and quick, so the edge only shows up over a large sample, after costs.

Disclaimer: This article is for educational purposes only and is not investment advice. The worked example uses hypothetical numbers, not real stock prices. Trading and investing involve risk; please do your own research or consult a SEBI-registered adviser before making any investment decision.