How to Create Strategy: One Dataset of One Market Type

One Dataset of One Market Type

First, traders should develop a profitable strategy (e.g., based on signals or specific/dynamic values of indicators, etc.) on one dataset of one market type. 

Simple = buy and hold and sell later

Normal = buy and when it reaches a peak sell, buy at pull back

Complex = buy, wait for peak, sell, when it reaches pull back buy, sell at the previous high, wait for bottom and buy, sell 20% below previous high etc.

Simple data = simple algorithm = can work on multiple datasets = low success = underfitting

Normal data = normal algorithm = best fit = optimal success for multiple datasets

Complex data = complex algorithm = high success on specific dataset = low success for any other dataset = overfitting

It is important that the testing dataset (where traders back-test their strategy) contains only that one specific market type, otherwise and in order to make the strategy profitable it would have to be over optimized, so-called overfitting, and as a result, it would work only with the specific dataset, and the strategy wouldn’t work on the selected market type, which is what is desired.

Time of one trade. It can be just one or three candlesticks, so called scalp, or one move in a trend, which can be tens of candlesticks. This depends on the strategy.

Fees and spread. Traders have to consider fees (and spread) in the strategy. For example, the trading fee can be ~0.1% for each trade (can be fixed such as 1$ or 10$ etc.), so a trader has to count ~0.2% for a buy and then sell. 

When to Enter

During the development of a trading strategy, traders should be looking for a confluence of multiple indicators that signal the same outcome in order to increase their chances. Especially, indicators from different categories (as we described before), such as trend, price level, and momentum.

An example buy condition can be:

  • The price is above EMA 200 indicating a long-term trend
  • Lagging span of Ichimoku cloud is away from the price indicating an active trend
  • The price is near previous high or high volume at current price level (PBV) indicating a price level where buyers can come in
  • Candlestick reversal pattern indicating a momentum

Before entering, traders should have figured out when to exit (take profits) and when to stop loss, which we cover in the next sections.

Adding to a loser

Is increasing position size in a losing trade (which is moving to the opposite direction). When making a strategy, consider adding to a loser to average down the entry price or take advantage of the lower price.

When to Exit and Take Profits 

Traders should identify levels of opposing pressure, such as support (when selling) / resistances (when buying), high volume levels, previous highs or lows, and they (traders) can take profits at these levels.

Note, if there are no levels or the price broke ATH - price discovery (all time high), Fibonacci extension tool can be used to define take profits.

As can be seen from the image above, the Fibonacci was drawn from low $44 000 USD to high $50 000 USD. The 1st level of $53 000 USD which is 1.414 fib extension, and the 2nd level fib extension 1.272 where the price reverted (1st) and reacted (2nd). 

It is possible to use these levels to define take profits.

When to Stop Loss

Traders can take the closest level, such as support (when buying) / resistances (when selling), high volume levels, previous highs or lows in the opposite direction they are trading, which gives them a price level where to set stop loss. 

In order to prevent being stopped (that your stop loss is hit and you end up in a loss and you are selling an asset that would potentially grow in price) by a stop-hunt, it is advised to move the stop loss a bit away from the level (for example, the amount that you move away from the stop loss can be a 20-period-long ATR), e.g. set the stop loss to a lower price that was originally planned (when buying).

ATR (average true range) gives an average volatility of an asset over a period of time.

Risk-reward-ratio (RRR) should be over the value of 1, which means that reward (the difference between take profit and entry price) should be higher than risk (the difference between entry price and stop loss). If RRR is below 1, it is statistically less probable that the strategy will be profitable if repeated.

Hard stops. Setting a hard stop can help to minimize losses. For example, if the trader loses more than 5% of their portfolio or $5k per day, he/she stops trading that day.

Risk-reward-ratio (RRR) relation

It is more probable that a trader hits tight stop loss than wide stop loss. Some strategies work with tight stop loss and some with wide and the trader should play with this variable when developing a strategy.

In general, setting tight stop-loss (high number of RRR) yields a low number of wins but the wins are of big size. Vice versa - setting a wide stop-loss (low number of RRR) results in a high number of wins and the wins are of small size.

Note also, having high win rates results in high fees.

In the case of uptrends, it is profitable to remove stop-loss and always count on the price to go up, eventually. This yields additional profit over holding (sometimes even double or triple) as traders buy in pull backs or near average price levels and sell with some defined profit. For example, traders can buy near EMA with p = 20 and RSI below 40 and sell only if they reach 1.2% profit.

BUT in case of a downtrend, traders can easily run out of money (because they are buying pull-backs) and they have to hold through the whole duration of the downtrend which does not yield any profit.