Mizar Research Series Vol. I: Candlestick Bars

Mizar
4 min readApr 30, 2021

--

Dear Mizarians,

Welcome to the first volume of the Mizar Research Series, where the Mizar team shares knowledge with their community to help them build better algo-trading strategies. Today’s article will be about candlestick bars, which are widely used in the industry to build trading strategies.

Time Bars

The most common candlestick bars are time bars, which are sampled based on time. Typical frequencies are (1min, 3min, 5min, 15min, 30min, 1h, 2h, 4h, 6h, 12h, 1D, 1W, and 1M).

Figure 1: Time bars

Since time bars simply sample based on a frequency, they exhibit an undesirable property, where they oversample when there is low trading activity and undersample when there is high trading activity. For example, between 12:30 April 22 and 01:00 April 23, the volume is much higher than between 02:00 April 24 and 14:30 April 24. So one could argue the amount of information in the former is higher than in the latter, but since they are sampled on time this will lead to undersampling and oversampling respectively.

Volume, Tick, and Dollar Bars

Volume, tick, and dollar bars are sample based on whether a certain amount of volume and dollars have been traded or when a certain amount of ticks have passed. Therefore, when there is high trading activity, these bars will sample more bars by design compared to the time bars. Another way to look at it is that these types of bars automatically zoom in and out depending on whether there is high or low trading activity respectively, while the time bars are fixed.

Figure 2: Volume bars
Figure 3: Tick bars
Figure 4: Dollar bars

Dynamic Volume, Tick, Dollar Bars

A good rule of thumb to set the threshold at which a bar is sampled is to look at the average daily volume, dollar volume, or the number of ticks and set it to 1/50th of that.

For cryptocurrencies fixing the threshold to a constant level can result in too many bars being too few bars, since the trading activity has changed drastically over time. Therefore, it is desirable to adjust the threshold dynamically, which is how the volume, dollar, and tick bars are created in Mizar. The threshold is adjusted daily based on 1/50th of the moving 30-day average. In Figure 4 this behavior can be seen live when we look at the volume, we see that the volume is constant, but it changes slightly after midnight, which reflects the dynamic update of the threshold.

Also, the bars in Mizar do not only reflect the open, high, low, close, and volume as is typical with candlestick bars. At Mizar, all tick data is ingested, and therefore it is possible to aggregate information based on buy and sell orders and count the number of trades. Bars created by Mizar reflect much more information than normal candlestick bars, which we hope will give our users a competitive edge when developing their strategies.

Conclusion

In this article, we described the different types of candlestick bars that are available in Mizar and how these bars differ from each other. In Mizar, we offer time (with different frequencies) and volume, dollar, and tick bars, whose thresholds are updated dynamically daily to ensure that the creation of bars is stable over time. Finally, all available bars reflect additional useful information which is not typically available in candlestick bars (distinctions between buy and sell, number of ticks).

Join Mizar Today!

Help us build the crypto trading platform you want to use.

Mizar is a next-gen trading platform where traders can search for strategies to securely allocate capital or get access to a wide range of trading tools to put their assets in motion. Mizar makes trading simpler and more accessible to everyone, subscription-free.

Be an active part of the MIZAR crypto trading community.
Sign up here or visit mizar.com

Discord | Telegram | Telegram Announcements | Twitter | Website | Medium | Youtube

--

--

Mizar
Mizar

No responses yet