I wrote a bot and gave it 25k to trade

Today, 6 months have passed since the system went into operation. A new high reached.
I am cautiously satisfied for now.

 

Realtime scan and trading are both performed on MNQ.
QQQ only used for backtests.

 

Out of curiosity, why not use MNQ for backtesting?

 

Because of the rollover problem. I get historical data directly from IB and I'm not willing to write code to cut and paste different contracts. I've had issues working with continuous contract data, so I use QQQ, which can be continuously downloaded several years back and matches MNQ almost identically in terms of % variations.

 

Makes perfect sense. I trade on a continuous MNQ ticker IB offers, so I'm quick to forget that its has expiration dates.

 

I think you took certain risks when you made the robot.

 

Meaning that trading would otherwise be a riskless activity?

 

A risk-free trading algo would look something like:

if $SPY < 0 then
buy next bar;
else
pass;

@Brightbreaker -- you're trading psychological risk (bad decision making) for some infrastructure risk (systems going down, code breaking). I find the latter more controllable.

 

Just don't try that with oil

 

Hi terminus,

how often do you resetting parameters?
I have a system that I re-optimize every 4 weeks, basically take profit, stop loss and trailing levels in order to adjust it to new market condition.
I am not feeling very confident with this methodology even I am profitable for the last 2 months.
Are you adopting this methodology too or what?

thank you