The evidence
The average year drops 14% and finishes up 13%.
Both numbers describe the same year, and both are averages over the same period. That pairing is the entire argument for buying dips, and almost nobody states it next to the argument against.
We are going to state both.
The average year contains a double-digit drawdown and still finishes higher. Dips are not what goes wrong with a bull market. They are what a bull market is made of.
The case against us
There is a famous study that says dip buying does not work
Buy the dip loses to dollar-cost averaging more than 70% of the time.
Nick Maggiulli, Even God Couldn’t Beat Dollar-Cost Averaging
Every 40-year period between 1920 and 1979. One investor puts $100 a month into the market. The other saves that $100 in cash and deploys it only at the exact bottom between all-time highs, with perfect foresight of where that bottom is.
Miss the bottom by two months and the strategy underperforms 97% of the time.
It is a good study. The finding is real, it is widely cited, and if you have ever wondered whether dip buying is a myth, this is almost certainly the thing that made you wonder.
The distinction
It tests a strategy that shares our name and nothing else
Read the method again. The losing strategy is sitting in cash waiting. It loses because it is uninvested while the market compounds, not because oversold conditions fail to bounce. That is a question about allocation. It is not a question about whether a pullback inside an uptrend resolves upward.
Hold cash. Wait for a bottom.
An allocation decision. It fails because the money sits out of the market for years at a time, not because pullbacks stop resolving upward. Being right about the bottom does not repay the years of compounding you skipped waiting for it.
Stay engaged. Buy oversold inside an uptrend.
A trade-level decision on a horizon measured in days. The question is not whether to be invested. It is whether a short-term oversold reading, in a market already trending up, produces a better than average return over the next few sessions.
Two different questions, one study. Conflating them is the most common mistake in this entire argument, and it gets made by people on both sides of it.
What the research supports
The 2-period RSI
The best-documented public example of the method: buy short-term oversold readings, but only while the longer trend is up. Published by Larry Connors and documented at StockCharts ChartSchool.
Take long signals only while price is above its 200-day simple moving average.
Buy when the 2-period RSI drops below 5 to 10. Lower readings historically produced better returns.
Close on a move back above the 5-day simple moving average.
Connors’ testing covered hundreds of thousands of trades.
And what it gets wrong
- The original rules use no stop loss. Connors found stops reduced returns, but the tradeoff is the risk of outsized losses.
- Signals are frequently early. Moves often continue after entry.
- The method is not designed to catch major tops or bottoms, and extended trends punish mean reversion repeatedly.
These are published because they are true, and because the failure mode of every dip-buying system is the same one: the dip that does not stop. Any vendor who shows you an equity curve without showing you this is selling you the curve.
How we test
A long-only backtest in a bull market is measuring the bull market
Any strategy that only ever buys will look good over a period where price mostly rose. That is not evidence of an edge, it is evidence of a rising market, and the two are almost impossible to tell apart unless you deliberately separate them. These are the rules we hold our own results to.
Benchmarked against buy-and-hold
Every result is shown beside the same instrument simply bought and held over the same window. The gap between the two is the only number that means anything. If there is no gap, the strategy did nothing and we will say so.
Time in market reported
A strategy invested 20% of the time that captures most of the return is doing something real. One invested 95% of the time is an index fund with extra steps. You cannot tell which you are looking at without this number, so it gets published.
Costs included
Commission and slippage modeled, not assumed away. Short-horizon mean reversion is exactly the kind of strategy that dies on transaction costs, and a backtest that ignores them is fiction.
Tested out of sample
Rules are fixed on one period and then tested on a period that was never used to tune them. Any threshold that has to be re-optimised to keep working was overfit, and we would rather find that here than in your account.
The bad periods shown
Results include 2000 to 2002, 2008, and 2022. A dip-buying strategy that has only been tested on a bull market has not been tested. The drawdowns get published at the same size as the returns.
None of this makes a strategy good. It makes a result honest, which is the part you cannot verify for yourself before buying.
The filter
We buy dips, not downtrends.
Every serious piece of mean-reversion research reaches the same conclusion from a different direction: the oversold reading is not the edge. The edge is the oversold reading inside an established uptrend. Strip the regime filter out and you are no longer buying a dip, you are catching a falling knife, and the historical record on falling knives is exactly as bad as it sounds.
The pullback interrupts an uptrend. It is a discount on something already working.
The identical pullback shape, inside a downtrend. This is where dip buyers get hurt.
Illustrative shapes, not price data. The point is that the pullback alone tells you nothing. Only the regime around it does.
One side only
We only trade one side, and that is where the edge comes from
Not a preference, and not a risk tolerance thing. The long side of a broad index is the only place in this business where the structure is already tilted your way before you do anything clever.
The long side has a tailwind you did not have to find
Equity indices drift up over time for reasons that have nothing to do with anyone being clever: a risk premium someone has to be paid to accept, earnings that compound, and index committees quietly removing the companies that stopped working. You are collecting something structural, not exploiting an inefficiency.
Structural edges do not decay when published
This whole site gives away the rules, which would be reckless if the edge were an inefficiency. It is not. An inefficiency dies the moment enough people trade it. A risk premium does not, because it is compensation for a risk that never goes away.
A symmetric system has no tailwind in either direction
Short edge has to be taken from another participant’s mistake, which puts you in a crowded, adversarial, decaying game. And you carry unbounded loss, borrow cost and squeeze risk while fighting the drift the entire way. Same effort, worse structure.
One side means half the decisions
You stop asking "long or short" and only ask "is this a good long". It does not make trading easy. It makes it simpler, which is a different and more useful thing, and every decision you do make is pointed downhill.
The chase
The strategy that works in every market is not available at your desk
Everyone goes through the phase. A system that takes both sides, handles every regime, needs no judgment, and never has a bad quarter. Weeks of parameter tuning. A backtest that finally looks perfect right before it meets the next quarter and does not survive it.
We did it too. The expensive part was not the losses, it was the eighteen months.
Here is the honest version: symmetric systems that genuinely work absolutely exist. They belong to firms with co-located hardware, a borrow desk, and a dozen people whose entire job is research. That is not a character flaw on your part, it is a resourcing fact, and no amount of optimising a retail backtest closes that gap.
What is available to you is the side that already tilts your way. One direction, one question, and a filter that tells you when to sit still.
Narrower is not a compromise. It is the only version of this that a person can actually run.
How often this happens
Dips are the normal operating condition
The rules are published. The tools run them.
Everything argued on this page is built into the indicator suite, with the regime filter first and the entry trigger second.
See the indicatorsSources
- Invesco: Stock market corrections and what investors should know
- Of Dollars And Data: Even God Couldn’t Beat Dollar-Cost Averaging
- StockCharts ChartSchool: RSI(2)
Nothing on this page is a claim about the performance of any We Buy Dips product, and nothing on it is investment advice. The statistics describe historical index behavior. Past market behavior does not predict future results, and every strategy discussed here has losing periods.