🪀 Mean reversion: betting on the snap back

🎯 In 30 seconds: Mean reversion bets that a price which has moved unusually far, unusually fast, will move partway back toward its recent average. It works best on short time frames, in liquid shares, in the direction of the longer trend. Pairs trading applies the idea to two related shares, but in India you cannot hold a short in shares overnight in the cash market, so the short leg needs stock futures or SLB. The big danger is the falling knife: a share that is cheap for a reason and keeps falling, sometimes through a lower price band where your stop cannot fill.

The idea in plain words

Watch the onion price at a vegetable market. A truck strike or a rumour can push it up sharply for a few days. Then supply catches up and the price drifts back toward normal. It rarely returns exactly to where it was, and sometimes the jump was the start of a real shortage. But on most days, extreme moves fade.

A mean reversion trader looks for the same thing in share prices: a move that is large compared with the share's normal behaviour and that has no new information behind it. The trader bets on a partial snap back, takes a small profit and gets out. It is the opposite mindset to momentum, and the next sections explain how both can be true.

Three common setups

1. Oversold bounce in an uptrend. A share above its rising 200-day average falls sharply for several days without bad news. You buy expecting a bounce toward its 20-day average. Indicators like RSI or Bollinger Bands are common ways to define "sharply"; see Momentum indicators and Volatility indicators.
2. Range trading. A share moves sideways between clear support and resistance for months. You buy near support and sell near resistance, with a stop just outside the range. See Support and resistance.
3. Pairs trading. Two closely related shares usually move together. When the gap between them becomes unusually wide, you buy the cheaper one and short the dearer one, betting the gap closes. Covered in detail below.

The intraday version, the gap fade, sells a share that opens far above yesterday's close on weak news, expecting part of the gap to fill. See Intraday trading.

Measuring "too far": the z-score

A z-score tells you how many standard deviations a price is from its average. The formula:

z = (price minus average) / standard deviation

If a share's 20-day average close is ₹600 and the standard deviation of those closes is ₹12, a price of ₹576 has z = (576 minus 600) / 12 = -2.0. A price of ₹570 has z = -2.5. Prices more than 2 standard deviations from the average are uncommon under normal conditions, which is why many traders use -2 as an "oversold" line. Bollinger Bands are simply lines drawn at the average plus and minus 2 standard deviations.

A z-score is only a description. It does not know about results, a SEBI order or a promoter selling shares. Always check the news before treating a low z-score as a bargain.

Worked example: an oversold bounce

Hypothetical. Konkan Cements Ltd is in a long uptrend above its rising 200-day average. With no company news, it falls five days in a row from ₹620 to ₹570 during a market-wide wobble. Its 20-day average is ₹600 and the standard deviation ₹12, so z is -2.5. On day six it closes up at ₹572.

If the target is hit: 117 x ₹28 = ₹3,276 gross. Delivery charges for the round trip (STT at 0.1% on both sides, stamp duty, exchange and SEBI fees, GST, a hypothetical ₹20 brokerage per order and a hypothetical DP charge) come to about ₹217, leaving about ₹3,059. If the stop is hit: about ₹1,989 plus ₹212 of charges, roughly ₹2,200.

Notice the modest target. Mean reversion trades usually aim for a partial snap back, not a new high. That means a higher win rate but smaller wins than trend following, and the charges are a bigger share of each win. On the trade above, charges took about 7% of the gross profit.

Pairs trading, Indian style

Pick two companies that earn their money the same way: two private banks of similar size, two cement makers in the same region, two IT services firms. Their share prices will not move in lockstep, but the ratio between them often stays in a band.

Hypothetical. Deccan Bank trades at ₹1,200 and Malabar Bank at ₹750, a ratio of 1.60. Over the past 60 days the ratio averaged 1.50 with a standard deviation of 0.04. So:

z = (1.60 minus 1.50) / 0.04 = 2.5

Deccan looks expensive relative to Malabar. The trade is: short Deccan, buy Malabar, in roughly equal rupee amounts, and close both when the ratio returns near 1.50. If the whole banking sector rises or falls, the two legs largely offset. You are betting only on the gap.

Now the Indian catch. A retail trader cannot keep a short position in shares overnight in the cash market. A multi-day pairs trade therefore needs one of these:

• Stock futures on both shares. Both must be in the F&O segment. Each contract is a fixed lot set by the exchange, usually worth several lakh rupees, so matching the two legs in rupees is rough. Both legs need margin, profits and losses are settled daily, and contracts must be closed or rolled before expiry. See Futures basics.
• Securities lending and borrowing (SLB). You borrow the shares through the exchange's SLB platform for a fee and sell them, then return them later. Availability and fees vary.
• An intraday pair. Both legs closed the same day. This avoids the overnight problem but leaves little time for the gap to close.

Hypothetical futures version: Deccan Bank's lot is 700 shares (700 x ₹1,200 = ₹8.4 lakh) and Malabar Bank's is 1,100 shares (1,100 x ₹750 = ₹8.25 lakh). The legs are close but not equal in rupees, and you cannot trade half a lot. With both positions at full futures size, a beginner's whole account can depend on one pair. That is why pairs trading belongs later in your learning, after Margin in India and Short selling in India.

Evidence: Gatev, Goetzmann and Rouwenhorst tested a simple pairs rule on US shares from 1962 to 1997 and found excess returns for the best-matched pairs, while noting that part of the profit may have come from trading effects such as the bounce between bid and ask prices. Their data is old and from another market. Treat it as evidence that the idea has worked somewhere, not that it works for you after Indian charges and futures margins.

What price bands do to the trade

Indian exchanges limit how far many shares can move in one day, and that changes mean reversion in two ways.

• Shares outside F&O have fixed daily price bands of 2%, 5%, 10% or 20% either way, set by the exchange. A share in a 5% band that falls 5% on bad news can lock at its lower band with only sellers in the order book. If you bought the "oversold" dip the day before, you may not be able to sell at any price until the band opens, and it may lock again the next day.
• Shares in F&O have no fixed band. The exchange applies a dynamic operating range (10% to start, widened in steps as price approaches it), so moves can be larger, but you can usually trade.
• The whole market can be halted if the Nifty 50 or Sensex falls 10%, 15% or 20% in a day under the market-wide circuit breaker rules.

A practical rule: run mean reversion trades in liquid, F&O-segment shares, and treat a share sitting at its lower band as a warning, not a discount. See The rules every Indian trader runs into.

Why mean reversion and momentum both exist

It sounds like a contradiction: momentum says winners keep winning, mean reversion says they pull back. Research suggests the answer is the time scale.

A rule many traders use: take mean reversion trades only in the direction of the longer trend. Buy short dips in shares that are in long-term uptrends. Do not buy dips in shares in long-term downtrends, which is where the next problem lives.

The falling knife

"Never catch a falling knife" is an old market saying. A share that has fallen 40% can fall another 40%. The z-score will say "very oversold" all the way down.

In Indian markets, sharp falls that keep going have often come with warning signs that a price chart cannot show. Before buying any big drop, check the company's recent exchange announcements on NSE or BSE for:

• an auditor resigning or qualifying the accounts,
• promoters pledging more shares, or lenders selling pledged shares,
• a SEBI or exchange order, or the share moving into a surveillance framework,
• a large debt default, a rating downgrade or a delayed results filing.

Any one of these is a reason to stay away, not a reason to buy cheaper. See Qualitative analysis: moats, promoters and red flags.

Try it

Choose five large shares in the F&O segment. In a spreadsheet, enter the last 60 daily closing prices of each from NSE's historical data. Add a 20-day average and a 20-day standard deviation, then a z-score column. Mark every day the z-score went below -2. For each, record what the price did over the next five trading days, and whether there was company news that day (check the exchange announcements). Then pick two banks from the same index and chart the ratio of their prices with its own 60-day z-score. Note how often the ratio crossed 2 or -2, and whether it came back.

Sources

• NSE: Price bands
• NSE: FAQs on price band and operating range flex
• NSE: Circuit breakers
• NSE: SEBI turnover fees, STT, stamp duty and other levies
• NBER: Gatev, Goetzmann and Rouwenhorst, Pairs trading: performance of a relative value arbitrage rule
• NBER: Lehmann, Fads, martingales, and market efficiency (short-term reversals)
• IIM Ahmedabad: Agarwalla, Jacob and Varma, Four factor model in Indian equities market (working paper, PDF)