Candlesticks are popular. They are very useful for visual displays, but our previous backtests indicate candlesticks might be useful to make trading strategies as well. In this article, we look at what is the best time frame for candlesticks:
The best time frame for candlesticks is daily bars and relatively short holding periods from 1 to ten days. Thus, candlesticks are most useful for short-term trading. We backtested different time frames from 15-minute bars to monthly bars.
Candlestick research (backtested)
Before we continue looking into to find out what is the best time frame for candlestick patterns, we’d like to take the opportunity to remind you about our very thorough research we have done on candlesticks and returns.
We quantified all the candlestick patterns we could find (75 patterns) and put them all into trading rules that are 100% testable. Thus, you can know with certainty which patterns are working and which ones don’t work. We backtested the patterns on S&P 500, but you can (of course) test on other assets. You can order or read more by clicking on the banner below:
What is the best time frame for candlesticks?
Most blogs and articles argue that candlestick patterns work well in all time frames. Is that really true? We don’t think so. Let’s do some backtests to find out what is working and what is not working.
In a previous article, we backtested 3 candlestick patterns that work. The patterns were tested one by one, but let’s test all 3 in one backtest (not 3 separate backtests). We recommend reading that article to better understand the background of this article.
The candlestick patterns in question are:
- Bearish Engulfing Pattern (yes, it says bearish, but it is more bullish than bearish)
- Three Outside Down Pattern
- Bullish Harami
Why did we pick these candlestick patterns? We picked these patterns based on the article named Do Candlesticks Work? In that article, we tested 23 different candlestick patterns and the 3 above were among those that returned the best results.
In this article, we test the above 3 candlestick patterns on the ES futures contract (S&P 500) and the ETF with the ticker code SPY. We enter a trade when one of the above 3 patterns gets a signal and we exit the trade when today’s close is above yesterday’s high. Commissions and slippage are not included.
Candlestick patterns using a daily time frame
The first test is done on daily bars. The 3 candlestick patterns above return the following equity curve on the ES-futures contract from 2011 and onwards (tested with 100% margin):
There are 159 trades, the average gain per trade is 0.42%, the win ratio is 72%, the profit factor is 2.5, the annual return (CAGR) is 6.68%, and the time spent in the market is only 19%. These numbers are pretty good for such “simple” trading rules.
Candlestick patterns using a weekly time frame
Let’s switch to weekly bars and see what happens (with the same criteria as used on daily bars):
As expected there are fewer trades but the result is pretty good:
75 candlestick patterns – historical backtests and performance
Unfortunately, the weekly test is fooling us. If we switch to a longer testing period from the inception in 1993 we get the following equity curve:
This is a pretty devastating equity curve and even after 20 years, we have not recovered from the dot-com bubble.
If we test SPY on daily charts by using the same period the result improves a lot:
Thus, we conclude that candlestick patterns work much better on daily bars than weekly bars.
Candlestick patterns using a monthly time frame
This is the equity chart on monthly bars from 1993:
As expected, there are very few trades. Additionally, it performs much worse than on daily bars: the average gain per trade is 3.2% and the CAGR is 3.1%.
Candlestick patterns using an intraday time frame
Do candlestick patterns work on intraday bars? Let’s return to the ES futures contract and test on intraday data.
First, we test by using hourly bars. The candlestick patterns are the same but we change the exit criteria: the last three bars must have a close that is higher than the previous high.
We get the following equity chart from 2011 and onwards:
There are plenty of trades: 338. The average gain is a respectable 0.32% and the CAGR is 10.5%.
Unfortunately, you spend 85% of the time in the market and thus this is closer to a buy and hold strategy than all the other previous time frames we tested.
If we switch to an even shorter time frame, 15 minutes, the results improve but then we risk giving away a lot of the profits to commissions and slippage:
The 967 trades generate 14.2% annual returns, but the time spent in the market increases to 89%.
Do candlestick patterns work on all financial assets?
This article has only focused on the S&P 500. This is not a coincidence. The stock market has a tailwind in the form of monetary inflation, productivity gains, and rising profits. Most other assets don’t have this tailwind, except for gold. For example, forex is a relative valuation between two currencies and is hence not a productive asset (like stocks).
We tested candlesticks on many assets like oil, gold, silver, Swiss Franc, Yen, bonds, etc. but candlesticks work best, by far, in the stock market.
Perhaps quite ironic, candlesticks don’t seem to work very well on Japanese stocks (ironic because the candlestick charting originated from Japan). The US stock market seems to be the best market to use candlestick patterns.
Recommended reading about candlesticks:
- Do candlesticks work? A quantitative test of 23 candlestick formations
- 3 bullish candlestick patterns that work
- How many candlestick patterns are there?
- Candlestick pattern – doji
- Candlestick guide
- Free trading strategies
- Monthly trading edges
If you want to have the Amibroker code for the 3 candlestick patterns plus 20 others, you can order it here (strategy 6).
Conclusion: What is the best time frame for candlesticks?
We tested many different time frames in this article to determine the best time frame for candlesticks.
Both long and short time frames didn’t produce any good equity curves. Long time frames might work but our backtests indicate large drawdowns.
Opposite, short intraday time frames are not very efficient either. For example, 15-minute bars involve noise and randomness and you need to employ wider exits to make them profitable, but then you are getting closer and closer to a buy and hold strategy.
So, what is the best time frame for candlesticks?
Based on the backtests we did we can safely conclude that candlesticks work best on a time frame of daily bars. The best holding period is from 1-10 days.