What types of trading algorithm exists ?

Any sorts of trading algorithms can be categorized in two broad category, A trend following system and reversal, note that a trend following does not mean some combination of certain indicators like MACD or RSI, definition of a trend following exists without any indicator, although a trend following system can be made using built in indicator. For most time trend following systems are used.

reversal signal

 

A one liner definition for trend following can ┬ábe a logic – whose nose is always ( or attempting) pointing to current price movement.

Beauty of trend following is that they always generate a signal when price based mathematical conditions are met, so you never miss any move and also due this, stop loss is built into the system, therefore money management is never an issue for trend system.

Downside of trend following is, that they suffer from significant lags, and if lags are tried to reduced, whipsaw problem starts surfacing. Usage of trend following system is suitable only for longer trade like swing trading or for investing.

Now coming to other type of signal system called Reversal, which is opposite to trend following system, they are not for everyone and you must have expert trading skills and swiftness, they can be used in intraday trading, reversal signal attempt to head-on with active trend, this is why they feel risky to many traders.

A reversal system usually gives buy/sell signal, towards the end of steep rally or fall, when slow churning moves happen they don’t generate signal. mostly you also need a stop loss to use them. reversal signal may provides you best price for entry when done with manual analysis they can be excellent day trading tool.


Regards
Abhishek K

Problems with Tick chart and their benefits

To many of us in our algorithm design phase, use of tick data frame might appeal magically as opposed to traditional minute charts, we might feel what if it can give better accuracy in signals !

While a tick data may certainly be helpful in machine learning (ANN algorithms) for the fact that candles they generate are smoother and gradual hence the training of neural network can be made easy, but they are also crippled by multiple challenges inherent to tick data.

Tick data

Tick candles are highly sensitive to liquidity – take for example two options (strikes) of same instrument applied with 200 tick grouping chart, so you will signals at different time for those two strikes, whereas the spot movement could be advancing with its own pace.

Candlestick patterns – candlesticks patterns generated by ticks are not reliable and not may not come at the places when you expect them. this is mainly because tick traders working on market may be using any arbitrary number of tick grouping (like 20, 100, 200). so there is lack of agreement, whereas in minute chart options are limited.

In the light of above context, I did not see much use case of using tick data, other than for the traders doing scalping in ultra short time in forex market.