Grid trading: Does it work?

[trendie]

I cant get past the slippage and spreads.
If I take a pair of buys/sells, and buy at 1.xxxx, the sell will be at 1.xxx3.
each pair will carry the spread, for example 3 pips, for each trade.
You then either take a spread-loss, eg 3 pips, or expect the trade to go a grid-size plus spread.

For example, if the spread is 3 pips, and the grid-size is 100 pips;
You either wait for price to move 100 pips for your win, but collect 97 pips due to spread,
OR wait for market to move 103 pips, and collect 100 pips.

Having buys/sells on the same account may be possible, but if not, you have the slippage due to different brokers giving potentially differing prices. You either ride out these kinks, or judge the situation, and try to ride out the trade movement, and try to squeeze out the spread differences.
For example, if you take a buy at 1.xxxx, you judge the move, and hold out to get your paired sell at 1.xxxx, by waiting for the price to move 3 pips so your trades are perfectly matched. If, however, you misjudge it, you may get the sell at a larger difference than the mere spread.
This merely adds discretion, and causes judgemental problems, which we're trying to remove.

Also, if the price is moving quickly, the buy/buy can be problematic also.
For example, if you're reaching the next grid-level, you have to take a profit from the previous levels trade, and add a new one.
Suppose, you had a buy at 1.xxxx, and price is reaching 1.x1xx.
you need to close out the buy at 1.x1xx, banking 100 pips, and at the same time trigger another buy at 1.x1xx. If price moves strongly, your new buy might be at 1.x1xz, where z is a couple of pips spread due to fast market moves.

I am still not accounting for requotes.

I am still going to find a way to make grids work.
(really weird, I am now seeking a technique to maximise the chop, and minimise the trend!)

EDIT: grids remind me of Point and Figure boxes!

 

[MrGecko]

I am still going to find a way to make grids work.

Dude

* commissions, spreads, financing
* finite account balance

make it impossible for grid trading to have a positive expectancy. Seriously, how is this not blindingly obvious?

 

[trendie]

It must be my trend-following nature, I keep introducing some notional bias, to use grids to only buy or only sell, which defeats the purpose somewhat.

the finite account balance problem can be mitigated by using large grid-size. But this is undermined by a much longer period of being potentially underwater, and profits being bled away through daily rollover costs.

Its the practicality of buy/sell spread and price differences across 2 different brokers, and the slippage of closing opening buy/buy or sell/sell in fast markets, also potentially causing getting skewed prices.

I think I may explore the Stochastics buy/sells idea, whereby I keep open all losing trades until I can nett them off (losing buy and sell pairs meeting at a nett null price at some retracement point, thus reducing margin requirements), whilst banking nett winning moves.
I much prefer the idea of opening new trades in direction of trend, a sort of pyramiding.

When a trend breaks, the idea of accumulating increasing liabilities seems counter-intuitive to me.

Still needs some thought though.

 

[counter_violent]

Ignore the nay sayers trendie and carry on.

Limit orders for both entry and exit take care of the slippage issues...any slippage therefore is positive not negative and goes some way (approx 50% ) towards the overall cost of financing positions.

Positions at regular increments are a no no. ODT mentioned the word "dynamic" probably the only sensible word he has ever used on t2w.

Two way grids also help to mitigate the drawdown element.

Small stakes and diversification across a range of instruments is key.

Forex is probably not the best group of instruments to consider as they can and do go on extended runs.

Each instrument must have a grid designed specifically.

You will need more than one account or sub accounts and be properly funded.

Regardless of the rigidity of grids, you still have another valuable weapon in the arsenal and that is the option of discretionary trades along the way This can be used to great effect and particularly using larger size stake in relation to the grid that's in play.

Grid trading is all about managing.

Sall for now.

 

[MrGecko]

ffs, do you actually want to make any money?

Or do you just want to believe that the key to making grid trading is right around the corner, then lay in bed at night thinking about how you will spend all the money you are going to make?

 

[counter_violent]

eeewwee

 

[trendie]

I am going to expound on this by taking a simple grid of 2 points:

A grid of 2 points. grid-size = G. (for examples sake, make this 100 pips,)

Start by opening a pair Buy/Sells at price X0.
The grid stops at point X1.
(assume for longs only for the time being. I am trying to show a generalised principle here)

How does it make or lose money?
If the market goes down, we're hedged. The buy cancels out the sell.

If the market goes up, we're moving towards point X1.
The position is hedged, as each pip gain in the long account is offset by the pip loss in the short account.

At reaching X1, 3 things take place;
1: the Buy is cashed in,
2: 100 pips are banked,
3: a new Buy/Sell is triggered.

You now have:
100 pips in the bank.
a shortfall of 100 pips in the short account. (nett =Zero)

You have locked in a losing position, since you have 1 Buy, and 2 Sells.

For the optimal result, what you want to see is the market return perfectly to point X0.

If the market returns to X0, the following 3 things happen:
1: the Short triggered at X1 is cashed in.
2: you bank 100 pips.
3: a new Buy is triggered. (NB: you dont trigger a Sell, since it was already open at start of experiment)

You have banked, in total 200 pips, have a Short at X1 showing -100 pips, and a matched Buy/Sell at X0. nett = +100 pips, since the Buy/Sell now cancel each other out.

If you decide never to close these positions, you simply have locked in an open loss of 100 pips no matter where the market goes. (balanced by 200 in the bank)

In this "perfect" scenario, as long as the market ranges between X0 and X1, you are fine and dandy.
Each cycle between X0 and X1 will bank 200 pips.
If you chose never to add any more grids nor take any further action, no matter how low price falls
or how high it rises, you simply carry the -100 open loss in perpetuity.
(margin requiremetns notwithstanding)


Caveat:
no guarantee that you will get the exact prices,
nor that you wont experience slippage,
nor that gaps wont skew the results.
If the grid-size G is larger than a daily size, you may experience overnight charges and commissions.


The X0 to X1 is the first part of the exercise. This is to understand the mechanics of a single simple grid.

 

[MrGecko]

omg

 

[trendie]

omg

lol

 

[trendie]

So, what happens in a "perfect" multiple grid?

Suppose X0 is now extended to X1, X2, X3 and X4?
Just as an example, you now have 4 grids.

In a perfect world, it would rise exactly 4 grid-sizes, and return to X0.

But, there is a difference. Lot numbers and margin.

Because, you have to initiate a new trade at each level.
At the start you open a Buy/Sell at X0.

I am not going to go into pedantic detail here, but in essence, in a clean move to X4;

You have banked 400 pips (the X0-X1 move, X1-X2 move, etc to X3-X4). Thats 400 pips.

However, since there has been no "wiggle" downwards, you haven't bacnked any sells, and theya re accumulating.
The Sell from X0-X4 is 400 pips underwater, the X1-X4 sell is 300 underwater, etc.
Total open losses are (400 + 300 + 200 + 100) = 1000 pips.
Total open Buys are (300 + 200 + 100) = 600 pips.
Considering the 400 banked, you are nett Zero.

Now should the move now make a perfect move back from X4 to X0;
you should bank 400 pips on the way down.
But, this time, you are accumulating the Buys as geometric losses, and accumulating linear sells gains.
But this is a mirror image of the up-move.
When you get to X0, you should have banked a total of 800 pips, have a nett floating loss of 400.

But, any further moves up and down from X0 to X4 results in 800 pips per cycle.

Note: the margin required here is the equivalent of 20 trades.
This is because you needed 10 lots ( 4 + 3 + 2 + 1) to keep the sells open.
You also needed enough margin to allow for 10 buys open on the way down.

Thats 20 lots (multiples of 100 pips) to bank 8 lots. (800 pips), plus a floating loss of 400 pips.
In this scenario of perfect moves between X0 and X4 and back.

 

[trendie]

What about range-breaks and trend-moves?

Well, I think the problem here is one of psychology.
The classic grid fails, not becasue grids may or may not work, but becasue they are built on the human failing of not taking losses.
Clearly, allowing a grid to increase in size indefintely, and thus accumulate geometric losses while collecting linear gains is unsustainable.

Whats the fix?
Dynamic grids, of course.

 

[trendie]

Lets take the X0 to X4 gris as an example.

Its perfection is based the assumption that the range has been perfectly tuned.
It might be.

What if it isnt?
Well, at the point X5 is triggered, you close out the most outlier Buy/Sells and take the 100 pip loss.
This re-dresses the grid to being a 4-level grid.

In the example, if the move is from X0 to X4, you have banked 400 pips, and accumulated losing sells.

If you close out the outermost sell, at a loss of 500 pips (X0-X5), and closed out the buy at a gain of 400 pips (X5-X1:remember the first buy from X0-X1 got banked), and add in the 100 pip profit from X4 to X5, you have moved the grid for no "notional" loss.

caveat: again, slippage, costs, etc may be invoked here.

This way, the grid will move with the trend until it starts to range again. While in trend mode, you simply make no profit, maybe some losses due to slippage.
But, once the range re-establishes itself, you go back to making profits on the wobbles between the new-found range.

This, I believe is the fix to avoiding accumulating losses in the classic grids. This is achieved by accepting some losses due to re-adjustment, rather than mindlessly sweating out losses due to some ingrained ego-driven need "never to take a loss".

 

[trendie]

Imperfect grids: the best kind!!

Now, in a notional 4-level grid, you make 8 wins, 4 levels on the "up" move, and a further 400 on the downmove.

Now, when it "wobbles, things get tasty.

When there is no wobble, you make 4 wins or 100 pips each.
If, however, it makes just one wobble, ie, falling back 100 pips, and then resuming its journey upwards, you bank an extra 200 pips.
eg, instead, of UP, UP, UP, UP. ( 4 wins)
you get UP,UP, DOWN, UP, UP, UP (6 wins)

any wobble from perfect means extra gains.

 

[trendie]

We can use ATR to determine the weekly range of a market, and set our grid-sizes to fit it.

eg, if Market X has a weekly range of 800 pips, you may decide that a 4-levels grid needs to be 200 pips apart.
You may decide to use a 5-level grid. If you wish to optimise to 800 pip,s the level would be 160 pips.

This way you can optimise to get the optimal grid-size, and even adapt it as markets change.

And when the market breaks into a trend, you dynamically move the gris by closing out buy/sell pairs until the grid settles.

 

[trendie]

note the margin required.

a 4-level grid needs you to have margin to keep (4+3+2+1) = 10 trades open.
(you dont need limitless margin since you're not stupid, and know how to trail the grid)

a 5-level grid requires (5+4+3+2+1) = 15 trades open.
Note, the more trades you have open, the more money you need. So be sensible about the grid-sizes and numbers of levels.

NB: Dont make the levels too big. Or, you run the risk of losing money through roll-over charges.

Do your own research over optimal grid-sizes, and time spent on open days, as each day may cost you.

 

[Shakone]

You're making things more complicated than they are. If you go long and short at one level, you hold no position and have paid the spread twice for no reason. If it rises 100 pips (where you say you'll cash in the long) and enter another short and long (if I understood correctly), then you now have 2 shorts and one long (this is what you posted), so your equity (including open trades) is the same except that now you've paid 4 spreads and have gained nothing, and are net 1 contract short from 100 pips up.

So you have paid 4 spreads to be net one contract short from 100 pips up. Now suppose I want to be net short from 100 pips up. I wait until it goes up 100 pips, and then I short. I have the same position as you, except I've paid 1 spread instead of 4.

Is this what you really mean? Because it is ridiculous. This is not what I would call grid trading

 

[trendie]

Have I beaten the grid?
What did I miss?
Any conceptual errors?
Workable?

 

[trendie]

You're making things more complicated than they are. If you go long and short at one level, you hold no position and have paid the spread twice for no reason. If it rises 100 pips (where you say you'll cash in the long) and enter another short and long (if I understood correctly), then you now have 2 shorts and one long (this is what you posted), so your equity (including open trades) is the same except that now you've paid 4 spreads and have gained nothing, and are net 1 contract short from 100 pips up.

So you have paid 4 spreads to be net one contract short from 100 pips up. Now suppose I want to be net short from 100 pips up. I wait until it goes up 100 pips, and then I short. I have the same position as you, except I've paid 1 spread instead of 4.

Is this what you really mean? This is not what I would call grid trading

I understand.
But I believe the grids are based on not having any bias.
If you "wait for price to go up 100 pips and then short" implies you know price wont continue up, and will fall. If it continues up, you're holding an increasing loss anyway.

Agreed, commissions are the problem, especially as you open multiple levels.
I was just trying to work out the mechanics of avoiding the exponential never-take-a-loss madness.

PS: "not grid-trading". I came across another grid-style, where you pick a price, and when price moves up, you trigger a sell, and further sells a set amount above this Point X, and trigger buys at intervals below Point X.
That is simply madness as its optimal condition is to return to Point X.
I have rejected that as a grid as its frickin' barkin', and such people deserve to lose their money.
(I need the market to tell me which direction to take, so would want to adapt to it)

 

[Shakone]

No. You're missing the point. At 100 pips away, whether it be 100 up or 100 down, I can get the same position as you, by paying 1 spread, that you pay 4 for. How your can your system ever make sense like that. If I mirror every position you take, and pay a spread of 2, whereas you're paying a spread of 8, who is going to come off better?

If you open two positions in the opposite direction from one level, you have paid the spread twice and have NO position. You have just given money away to your broker. If you want to give your money away freely, give it to a good cause.

 

[counter_violent]

LOL ...well trendie, you did'nt read all this in a book.

Good to see some original thoughts going into this. Well done.