Larry Connors built the 2-period RSI strategy to buy after a fall, inside a trend that is still up. StockCharts describes it as "a fairly simple mean-reversion trading strategy designed to buy or sell securities after a corrective period", and states its limit explicitly: "It is not designed to identify major tops or bottoms."
The RSI 2 trading system is not a bottom-picking method. It is a short-hold method that takes one leg of a decline inside an uptrend and exits when the stock returns to its own recent average.
This article uses Connors' two core conditions, a 10% stop and the 5-day SMA exit. The stop is this article's addition, and the section on stops below explains why that is a real departure from the original.
What the RSI 2 trading system measures
The RSI, developed by Welles Wilder, runs between 0 and 100. Its default setting is 14 periods, which is slow enough to keep readings inside the middle of the range for weeks at a time.
At 2 periods, the calculation reads only the last two bars of gains and losses, so it swings to its extremes often. That is the point. StockCharts notes that short-term traders sometimes use a 2-period RSI and look for readings above 80 and below 20, which sit further from the middle than the standard 70 and 30.
Connors' threshold is stricter. Buy signals come when "2-period RSI moves below 10, which is considered deeply oversold", and short signals when it moves above 90.
The rules Connors set out
StockCharts sets the strategy out in four steps.
Step 1: the trend filter. Connors uses the 200-day moving average. "The long-term trend is up when a security is above its 200-day SMA and down when a security is below its 200-day SMA." Long trades are taken only above it.
Step 2: the RSI level. Connors tested RSI levels between 0 and 10 for buying and between 90 and 100 for selling, using closing prices. The finding, in the source's words: "He found that returns were higher when buying on an RSI dip below 5 than on one below 10." The lower the threshold, the fewer the signals and the deeper the fall you are buying.
Step 3: the timing. You can act just before the close or on the next open. Connors' stated preference is before the close. The trade-off is stated plainly: buying before the close leaves you exposed to the next open, which may gap either way, while waiting for the open gives you more flexibility over the entry price.
Step 4: the exit. Connors exits long positions on a move above the 5-day SMA. StockCharts calls this "clearly a short-term trading strategy that will produce quick exits".
Adding a pullback condition
This article adds a third entry condition: price must close below its 5-day SMA before you buy.
That is not one of Connors' four steps. In his version the 5-day SMA is the exit benchmark, not an entry filter. Adding it as an entry condition does two things. It removes setups where RSI(2) fell below 10 while price was still above its short-term average, and it reduces the number of trades. Whether that trade-off suits you is a decision, not a rule handed down from the original.
This article keeps it, and labels it for what it is.
Where the stop goes, and the arithmetic it forces
Connors did not use stops. StockCharts reports that his quantitative testing, across hundreds of thousands of trades, found that stops hurt performance on stocks and stock indices. The page is equally direct about the consequence: "not using stops can result in outsized losses and large drawdowns. It is a risky proposition."
Its own conclusion is that a trader should still build an exit and stop plan for any system, and this article follows that. It uses a stop 10% below the entry price.
Now run the arithmetic that stop creates.
The exit is a move above the 5-day SMA, which by design happens soon and close to the entry price. Call the win $1.20 per share on a $43.60 entry, a 2.8% move. With a 10% stop, the risk per share is $4.36, so that win is 1.20 / 4.36 = 0.28R.
At 0.28R per win and 1R per loss, the breakeven win rate solves 0.28w = 1 - w, which gives w = 1 / 1.28 = 78.1%. A system built on small quick wins and a wide stop needs a high strike rate to stay level.
Two ways to change that number. Take the stop closer to the entry, which raises R per win and also raises the chance of being stopped out of a trade that would have reverted. Or hold for a larger move, which is the next section.
Exits and the holding period
The 5-day SMA exit. The one Connors uses. It fires when price closes back above its own five-day average.
A 10-day SMA exit. A looser version that keeps the position through more of the bounce. This article treats it as a variation, not Connors' rule. It holds longer and produces a larger move per win, which raises the R figure above and lowers the breakeven win rate with it.
A time exit. Close the trade after a fixed number of days whether or not either average has been crossed. Mean reversion that has not happened in a week is not the trade you entered.
One caution from the source about the signals themselves: "the RSI(2) strategy can be early because the existing moves often continue after the signal." In the source's Apple example from 2011, there were at least ten buy signals. It reports that losses on the first five would have been hard to avoid, because the stock zigzagged lower for months. One filter the page suggests for that problem is waiting for RSI(2) to climb back above 50 before entering. That trades a worse entry price for evidence that the fall has stopped.
A worked RSI 2 example
This example is hypothetical. The company, the prices and the outcome are invented to show the arithmetic.
A stock trades at $45.00 with its 200-day SMA at $40.00, so the trend filter passes. Over two days it falls to $43.50, below its 5-day SMA of $44.20, and RSI(2) closes at 7.2. Both core conditions and the pullback condition hold.
- Entry, on the next open: $43.60
- Stop, 10% below: 43.60 x 0.90 = $39.24
- Risk per share: 43.60 - 39.24 = $4.36
- Share count on a $50,000 account at 1% risk: $500 / $4.36 = 114 shares
- Position value: 114 x $43.60 = $4,970, which is 9.9% of the account
Two days later the stock closes at $44.80, above its 5-day SMA, and the exit fires.
- Gain per share: 44.80 - 43.60 = $1.20, a 2.8% move
- Gain on the position: 114 x $1.20 = $136.80
- In R terms: 1.20 / 4.36 = 0.28R
A 2.8% move in two days reads well. A gain of $136.80 against $500 of risk reads differently. Both are the same trade. This is why the position size and the stop have to be set together rather than one at a time.
Tracking RSI 2 trades in Swingfolio
Swingfolio ships this setup as a template, with parameters of its own. On the Strategies page, the "Import from Library" button opens a sheet titled "Strategy Library", and one entry is named "RSI-2 Connors Setup".
Hover a row and press Preview, and the modal lists what it creates. Entry Rules (3): "Price is above 200-day SMA (confirmed uptrend)", "RSI (2) is below 10 (deeply oversold)" and "Average volume above 500,000 (liquid stocks)", the last carrying an "Optional" badge. Exit Rules (2): "RSI (2) crosses above 70 (mean reverted)" and "Price hits stop loss". Under Risk Management: Stop Loss 3%, Target R-Multiple 1.5R, Position Size 2% and Max Holding 5 days.
Read those against this article before you import. The template exits on RSI(2) crossing above 70 rather than on the 5-day SMA, and its stop is 3% rather than 10%. Both are editable after import, and the difference is worth thinking about rather than clicking past. A 3% stop on a $43.60 entry puts the risk per share at $1.31, so the same $1.20 win becomes 0.92R instead of 0.28R. The breakeven win rate falls with it, from 78.1% to 1 / 1.92 = 52.1%.
On the trade form, a "Strategy" selector attaches the trade. Further down, opening the Pre-Trade Checklist shows that strategy's entry rules as a checklist. Tick the RSI rule when the reading was below 10 and leave it unticked when it was 12, because that distinction is the whole question this system asks.
The same form shows Exp. Risk, Exp. Reward, R:R Ratio and Port. Risk, with a warning reading "Risk exceeds 5% of portfolio value" past that line. It does not calculate a share count. The position size calculator takes an entry price, a stop price, an account value and a risk percentage. It prints the share count under "Your Position Size", with Investment Amount, Risk Amount and Stop Loss Distance below it. Add a target price and it adds Potential Reward and Risk:Reward Ratio.
Once the trades close, "P&L by Setup Type" on the analytics page groups them by the strategy attached to each and charts win rate and profit factor for every strategy side by side. That pairing is the one to watch here: a mean reversion system can carry a high win rate and a thin profit factor at the same time, and the card shows both. "Profit Factor by Compliance" splits the same trades into Compliant and Non-Compliant bars, which is what the ticked checklist feeds.
A separate "Sector P&L" card breaks results down by sector. It reads every closed trade in the account rather than only the ones on this strategy, so it answers a portfolio question, not a per-strategy one.
Common mistakes with the RSI 2 trading system
Buying below the 200-day SMA. Below it, an oversold reading sits inside a downtrend and the average the price would revert to is falling. The trend filter is the first of Connors' four steps for that reason.
Entering at RSI(2) of 12. The threshold is a threshold. Connors tested the range and found the deeper dips produced the better results in his testing, which is an argument for tightening the level rather than loosening it.
Holding through earnings. StockCharts notes in the same write-up that gaps "can wreak havoc on trades" and that three of the gaps in its example "occurred during earnings season". A gap opens past your stop rather than at it.
Reading a percentage gain as a good trade. A 2.8% win on a 10% stop is 0.28R. The percentage tells you what the price did. The R figure tells you what the trade did.
Take your last five mean reversion trades, divide each result by the risk you had on it, and write down the five R figures. If they average below 0.5R, the stop and the exit are not set for each other. Import the RSI-2 Connors Setup template so the next one records whether the RSI rule was met. Start the 30-day trial.