The harami pattern can be sketched as follows:
As we can see the second candle is completely engulfed by the first candle. The colors of the candles are not specified but according to some traders, the best combinations are if the first candle is bullish. Whether it is true or not, we will check below.
At first, however, let us have a look at a general pattern, i.e. we do not consider the color of the candles. The testing set-up is similar to this shown e.g. here and the figure below shows an example of a successful trade:
So here appear the results of backtesting on the Russell 3000 stocks, assuming that we enter on open of the next day after the hammer pattern (we do not wait for any bullish action afterwards):
|Total net profit||$264087|
|Number of trades||5155|
|Average time in market||23 days|
The results show that the strategy is at least profitable. To be honest, I think that there exist better strategies but the harami pattern occurs very often (note the huge number of trades) and therefore we cannot expect too much.
Let us repeat the backtesting, but this time we enter “tomorrow”, i.e. “today” the stock has to close above the second candle of the harami pattern. Some traders would say that it is better to wait until the price closes above the first candle but this will reduce the number of trades significantly.
Here are results:
|Total net profit||$109923|
|Number of trades||1105|
|Average time in market||29 days|
As we observe, the performance is better (see e.g. the higher profit factor and the lower maximum drawdown).
And now, as mentioned previously, we can do the same for the case where the first candle is bullish. So for the case when we enter just after a harami pattern:
|Total net profit||$62324|
|Number of trades||1348|
|Average time in market||19 days|
The only improvement is the lower maximum drawdown (perhaps due to the lower time in market, that is, the risk is also lower). The other parameters are virtually the same as before. We can repeat our backtesting for the case when we enter a day later (if the price closes above the second candle of the harami pattern):
|Total net profit||$15391|
|Number of trades||245|
|Average time in market||25 days|
Again, the performance is not really better than for the “standard” case.
Black-black or black-white – should be bad
Scenario 2, nback = 1:
but not worse than previously
Nback = 0;
Kelly 0.06, as bad as before