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 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.