Ok, I hope this leads somewhere...
Here's what I did.
Instead of intellectualizing it, I did some random examples like the one with the boss and the planes.
So I generated 2 sets of 10 random prices.
To keep it simple, just 10 random prices between 1.001 to 1.010.
Then I put a list of random long and short orders to each of these 10 prices.
I closed out the first order at the second price, and opened a new order at the second price, and closed it out at the third price etc...
Since choice of whether to go long or short was not within our control, all we had to play with was position size.
So I started with just assuming trading:
1 standard lot for all trades,
trading 1/2 a standard lot for all trades,
starting with 1 standard lot and cutting the size in 1/2 for every loss, and doubling the size for every win
starting with 1 standard lot and doubling the size for every loss and cutting the lot size by half for every win.
The first set of 10 random orders and prices, these being the total dollars generated for each group:
Test 1:
1 std lot -200
1/2 std lot -100
cut size on losses -262.5
inc size on losses 2700
Test 2:
1 std lot 2100
1/2 std lot 1050
cut size on losses 1700
inc size on losses 3500
So 1 lot vs 1/2 a lot, is as expected, 1/2 the gain/loss, so we'll just look at the standard, and adjusted lot sizes.
On the first test, cutting the loss size resulted in a bigger loss than just sticking with 1 standard lot. While increasing size on losses made a huge profit.
On test 2, cutting the size on losses, resulted in a smaller gain than just sticking with a standard lot, and again, increasing the size on losses resulted in the biggest gain.
That's obviously from only a small test, with no other restraints.
I don't know if this is what we wanted to start looking it... but it looks like, increasing how much you trade when you take losses works with random entries and prices.
Here's what I did.
Instead of intellectualizing it, I did some random examples like the one with the boss and the planes.
So I generated 2 sets of 10 random prices.
To keep it simple, just 10 random prices between 1.001 to 1.010.
Then I put a list of random long and short orders to each of these 10 prices.
I closed out the first order at the second price, and opened a new order at the second price, and closed it out at the third price etc...
Since choice of whether to go long or short was not within our control, all we had to play with was position size.
So I started with just assuming trading:
1 standard lot for all trades,
trading 1/2 a standard lot for all trades,
starting with 1 standard lot and cutting the size in 1/2 for every loss, and doubling the size for every win
starting with 1 standard lot and doubling the size for every loss and cutting the lot size by half for every win.
The first set of 10 random orders and prices, these being the total dollars generated for each group:
Test 1:
1 std lot -200
1/2 std lot -100
cut size on losses -262.5
inc size on losses 2700
Test 2:
1 std lot 2100
1/2 std lot 1050
cut size on losses 1700
inc size on losses 3500
So 1 lot vs 1/2 a lot, is as expected, 1/2 the gain/loss, so we'll just look at the standard, and adjusted lot sizes.
On the first test, cutting the loss size resulted in a bigger loss than just sticking with 1 standard lot. While increasing size on losses made a huge profit.
On test 2, cutting the size on losses, resulted in a smaller gain than just sticking with a standard lot, and again, increasing the size on losses resulted in the biggest gain.
That's obviously from only a small test, with no other restraints.
I don't know if this is what we wanted to start looking it... but it looks like, increasing how much you trade when you take losses works with random entries and prices.
