vince-lynch / busd-usdt-tradingbot Goto Github PK

Write pinescript to Backtest this strategy on tradingview.

Could be replaced by

Run script checks when prices seen (max, min) change.
OR
When trade is filled.

Currently working on the idea never to lose money. and look at opportunities regarding leverage etc once the code is solid and never loses,

(and also never stays in position for hours 'bagholding')

Because 'bagholding, with leverage could be a problem.

It spends so much time trying to break even, that when the price moves it can't adjust again. sometimes its hard to comprehend but if you watch the trades you can see it happening as you look at the order book.

There are some immediate actions we can take.

• Do not adjust sell to break even if asks are less than 25% of bids.
• Do not adjust buy to max if bids are less than 25% of asks.

Create a leveraged version of the strategy.

Slightly doing this just to force the writing of precise code, and to really think about how it works, whilst leaving open the opportunity for shorting aswell as long.

[Refactor] put different categories of method into folders

Since starting the project, haven't done a refactor, this would be the first step.

[11:53, 10/05/2021] Vince: it spends its entire life at breakeven
[11:53, 10/05/2021] Vince: The price, moves about the same length of time it takes to get a order filled
[11:54, 10/05/2021] Vince: Its quite a complex issue, and I think there might be many aspects to it. obvs its because its sooo liquid.
[11:55, 10/05/2021] Vince: so the market makers, are timing it perfectly. if it has to adjust the order to step down, then its already late, and it doesn't have the old order that was nearly through.
[11:55, 10/05/2021] Vince: I'm still unsure whether shorting fully addresses this issue, I think it might be that it just happens also the other way around (I could be wrong) because its the speed, timing, of orders, and which direction its going.
[11:56, 10/05/2021] Vince: Simple orderbook depth of market rules might help:

• do not adjust sell to break even if asks are less than 40% of bids.
• do not adjust buy to max, if bids are less than 40% of asks.
  [11:58, 10/05/2021] Vince: - Remove the 30seconds, and instead rely on immediate trade price, been a new min or max, as the trigger for the adjustment, or the immediate effect of an order been filled (This wont exactly make this high frequency, but still, its a start) - in addition, remove all unnessary external checks to carry out the transaction. - write this section in a faster/compiled language. .. like I say, this is no attempt to become high frequency, if we are doing that.. we are probably getting alot more than that wrong.
  [11:59, 10/05/2021] Vince: The issue with the short-side is, Ok, its great. but if there are fundamental problems with speed-timing, then they need to be addressed, and cutting the compounding in half might not help, unless it doubles the speed of getting filled AND increases the number of wins.

Store state.

could be database, could be json files.

could be useful for

• trades
• orders
• current position
• profit and loss
• charts? (p/l etc) to see times when not trading? etc.

What to do about downtrends
i.e. it could be several hours where all the bot tries to do is break even, this is wasted time.

Use cross account for trading instead of spot trading, to give us the flexibility of long/short without leverage.

Acceptance criteria

• gets details of cross account #18
• can trade long using cross account #19
• can trade short side using cross account (simultaneously)

We've not traded for 3 hours
because the software always insists on trying to break even.

And then the price might move further down.

This means the software misses hours of trading.. (Today it missed about 7hrs so far).

Its better to sell at a slight loss, and have 50 trades inbetween 2 levels.

For example, at the moment its trended between 90 - 91, or 91 - 92 for 4 hours.

So that's 6+ easy wins.! that we didnt get because we not taking a loss

Add CSV output of trades

• Basic init of some kind of P/L (Can be done in gsheets/excel from the output)
• basic init of a formula to calculate profit and loss in the project.

Cancel orders not going through, should not adjust an order if it hasn't cancelled first.