Selecting markets to trade

Goods

⌘K
  2. Home
  4. Docs
  6. Goods
  8. Statistical Arbitrage
  10. Selecting markets to trade

Selecting markets to trade

We will select paired markets for real trading among those trading instruments that have been tested on historical data, by analyzing the results of the profitability chart as well as statistical metrics. First, let’s remember which markets we tested.

On this small piece of the correlation matrix, we selected 38 pairs of markets that satisfy our filter for the correlation value to be greater than 0.85. We tested all these pairs of markets on history for 1 year, and now we will try to analyze the profitability graphs of these tests, both positive and negative, in order to learn how to identify markets with high cointegration.

BAND-CELO | RESULT: SUCCESS
PROFIT: +239% HIGH | FACTOR: 2.01 HIGH | CHART: PERFECT
AUDIO-CTK | RESULT: FAILURE
PROFIT: +8.7% LOW | FACTOR: 1.09 LOW | CHART: BROKEN
BAT-C89 | RESULT: SUCCESS
PROFIT: +105.4% HIGH | FACTOR: 1.55 MIDDLE | CHART: PERFECT
BICO-COMP | RESULT: FAILURE
PROFIT: +11.5% LOW | FACTOR: 1.19 LOW | CHART: BROKEN
BAND-C89 | RESULT: FAILURE
PROFIT: +92.1% MIDDLE | FACTOR: 1.46 MIDDLE | CHART: GOOD
BICO-CFX | RESULT: FAILURE
PROFIT: +4.7% LOW | FACTOR: 1.08 LOW | CHART: BROKEN
AUDIO-CELR | RESULT: SUCCESS
PROFIT: +144.2% HIGH | FACTOR: 1.61 MIDDLE | CHART: PERFECT
AXS-CAKE | RESULT: FAILURE
PROFIT: +27.1% MIDDLE | FACTOR: 1.68 MIDDLE | CHART: TOLERABLE
AXS-CELO | RESULT: SUCCESS
PROFIT: +148.8% HIGH | FACTOR: 1.77 HIGH | CHART: PERFECT
BALU-CRV | RESULT: FAILURE
PROFIT: +32.2% MIDDLE | FACTOR: 1.26 LOW | CHART: BROKEN
BAT-CELO | RESULT: SUCCESS
PROFIT: +166.4% HIGH | FACTOR: 1.82 HIGH | CHART: PERFECT
BAND-COTI | RESULT: FAILURE
PROFIT: +31.9% MIDDLE | FACTOR: 1.18 LOW | CHART: BROKEN
AUDIO-COTI | RESULT: FAILURE
PROFIT: +108.9% HIGH | FACTOR: 1.43 MIDDLE | CHART: BROKEN
Previous slide
Next slide

The conditions that the profitability chart of a pair of markets that we take into real trading must meet:

  1. The profit for the year at a risk of 10% of the deposit per transaction should be more than 100%
  2. Profit factor is more than 1.5
  3. The profitability chart does not contain strong drawdowns or sharp profits
By following these three simple rules, you can find those market pairs that have very strong cointegration; no ADF or Johansen test has such accuracy.
 
Now we have analyzed a random section of the correlation matrix of 255 elements, from which we have identified 38 elements suitable for testing, and then, using the analysis of backtesting results, we have found 5 pairs of markets that can be taken into real trading. However, the correlation matrix of the Binance exchange alone has over 38,500 market pairs, and probably contains 750+ pairs suitable for trading. You should find a few of these markets yourself and try them out in real trading. 

How can we help?