Introduction to grid trading system – Part II
As you have seen in Part 1 of this series, gridlocking could become a problem of trending grid system. Two possible ways of avoiding gridlocking are either having a different volume scheme or having different distances between the orders.
Having a different volume scheme is a problem itself. You may ask yourself what do you want to achieve with it: would you accept a slow grid scheme, moderate drawdown and cleanup time ? Or would you rather risking to destroy your account, but have a grid that cleans up fast ?
If you choose the slow version, you might consider laddering.
In this version you will introduce ladders, that will be at higher distances than the beginning levels of the grid, and with a higher lot size, in order to recover quickly the losses done by the other levels. However, the distances will be higher, and the waiting time would be moderate: faster than a gridlocked simple grid, but slower than a more aggressive scheme.
There would be an option to continue overlaying the small grid over the large grid, and setting the condition to achieve an overall target equal to a channel on the highest ladder touched. When that will happen, overlayed one-level grids that have a small lot size will be very deep in profit, so profits would’t be quite small. However, due to gridlocks over the small subsequent grids, it would take a long time for higher grids to get the system out of gridlock and achieve profits on all types of grid. This type of grid is advantaged by long trends, but disadvantaged by brutal swings followed by lower volatility.
If you choose the fast version, then what you have to do is to implement a Martingale-like lotsizing policy. The fast version would take as a risk a continuous flip : from first buy to first sell, then second buy, second sell and so on. The system would close the grid with profit any time the market would do two moves in the same direction. Like from first buy to second buy, or from second sell to third sell and so on. Given that loss progression is 1, 1+3, 1+3+5 and so on, – that, as a count of lost channels, because we will have a larger lot size per each level – we need to make sure that a won channel, at each level, is able to recover all previous losses. Therefore, we will use an algorithm to return the following sequence of lot multipliers: 1, 2, 8, 48, 384, 3840… Obviously lot 1 would get a profit of one channel, 2 for the second level would cover the loss of one channel, and insure the same one channel profit, 8 would cover 1+2 x 3 channels = 7 channels lost and one profit obtained. Should a 4th swing happen, the total loss would be 1+ 2×3 + 8×5 = 47 channels. Obviously it wouldn’t be adviced to increase losses beyond the third swing. This is why the size of the channel has to be pretty large, and has to be established by checking first the daily chart, so that the risk of having three swings in a row will be small.
In the next article we’ll talk about ranging grids.