Yesterday the stock market opened higher, climbed even more and then started sliding, only to end the day down and down more than 1.5% from the intraday highs it had reached just a few hours earlier. That’s as disappointing as days can get for people who own stocks, other than a crash of course. Here is how the market action looked like for the S&P 500:
Social media were full of messages how bearish this kind of market action is, as if people had been waiting for this. And yes, the logic seems to make sense: if the market cannot hold on to early gains, then sells off more than 1.5% intraday, it can hardly be seen as a sign of strength. The problem is: when was the last time the market was logical?
No matter how logical a trading concept may appear to be, and no matter how many experienced traders are telling you about it, it is always important to test. And we have plenty of stock market history to verify what really happens after this kind of disappointing days. So, that’s what I did. Here are the results.
I defined a disappointing day as a day on which the market opens above the previous day’s close, but ends the day below it. And the market must be down 1.5% or more from the intraday highs it reached to qualify as a disappointing day. To test what typically happens next we then look where the market was 5 and 20 trading days later, compared to the closing price on the disappointing day as defined above.
For the S&P 500 I found 143 of those days since the early 1980s. In 54.5% of the cases the market ended up being higher 5 days later, and in 57% of the cases the market was higher 20 trading days later (20 trading days is about 1 month). The average expectation is slightly positive, both after 5 and after 20 days. In 8 cases the market was down more than 6% after 5 days, and in 6 cases it was up more than 6% after 5 days. In 25 cases the S&P was up between 3 and 6% in the next 5 days, and in 47 cases the market climbed between 0 and 3%. Here is the complete distribution chart:
Looking where the market went after one month. In 20 cases (14%) the market was down more than 6% after 20 trading days and in 23 cases (16%) the market was up more than 6%. This chart shows all results for 20 days later:
I did the same test on the Nasdaq since the 1980s, which gives us a larger sample of 385 cases thanks to Nasdaq being more volatile. Also here we find that in 57% of the cases the Nasdaq is up 5 and 20 days after a disappointing day. In these 385 cases the Nasdaq has on average climbed 0.5% within 5 days, which is over 26% annualized. This means that a disappointing day like the one we have seen yesterday has historically been a bullish sign, not a bearish one. The average gain 5 days later is 3.5% and the average loss 5 days later is 3.55%. But there is a gain 57% of the time, so going short after a disappointing day has not been a profitable strategy in the last 30 years.
Here is what happened 5 days after a disappointing day in the Nasdaq:
In 26 cases the Nasdaq was down more than 6% after 5 days, in 37 cases it was up more than 6%
Here is what happened 20 days after a disappointing day in the Nasdaq:
In 44 cases (11%) we see the Nasdaq more than 10% lower after 20 days. In 58 cases (15%) the Nasdaq is more than 10% higher after 20 days.
Bottom line: a disappointing day in the market is more or less neutral with 57% odds that the market will be higher one week and one month after this day. The idea that this kind of days are a bearish omen is thus not supported by history. Slightly more often than not it is actually very bullish going forward. If there is anything bearish about these disappointing days then it will take additional qualifiers to get it to work. If you have ideas or links to similar studies then feel welcome to post them as a comment.
However, I got a feeling that in most cases after such a day down the price may have gone further down within the next 3 days. Interesting analysis, by the way, I am probably too lazy for this kind of study. Or It’s because I think that since I have access to 24 charts only it’s impossible to tell from those candles when the cash market opened really. Anyway, Tuesday price action was fast and sleek, with big reactions up offering reversal signals, while Wednesday, which used to be a more trendy day than Tuesday, was a choppy descending range for most of the day : if one trades those days similar way, there will be no two winning days.
All roads lead to Rome – all pauses and pullbacks resolve to the upside in a bull market.
In the 143 cases for the S&P there is indeed a negative expectation on the next day, so more often than not it dips a bit deeper still. But from 2nd day onwards the average expectation is positive.
In the 385 cases in the Nasdaq sample the 1st day after a disappointing day has a neutral expectation, on average no gain or loss on that day. Also here from 2nd day onwards the average performance is higher from the close on disappointing day.
So your intuition is not without merit.
So now we have the third day, which is up :) The lows have been rejected, Nasdaq tested below December lows so it looks like the market is reversing, BH on Monday should help.
I meant in my previous comment ’24h charts’ – where daily candles open at the close of US market or at midnight in Europe. I believe American brokers do offer charts which show candles only from cash open to cash close.
I recall I was doing in the past ‘strange’ statistical analysis of the market, mainly DJIA, for example what happened after a big Wednesday or big Thursday, but never found anythig which could lead to a set up. But from those studies something remain in my mind : I know that first 2 decades of the month are usually more trendy than the last, I know that the market often reverses at end of month, week or BH, that trend days happen mainly on Wednesdays and Thursdays, that Monday can be really slow, that the activity dies out before long weekends and so on. Nothing I got written down in my trading plan, these are subjective guidelines about what the market can do, but still I know first of all that the market may do whatever it likes regardless of rules. Have a nice weekend!
Thanks. Yes, there are edges based on day of the week, start of the month, and so on.. just like you describe. Some algos try to take advantage of those tendencies.