Quote from: PuertoLibre on January 24, 2014, 02:21:43 AM

Quote from: fible1 on January 23, 2014, 10:02:14 PM

Quote from: PuertoLibre on January 23, 2014, 01:12:06 AM

Quote from: PuertoLibre on January 20, 2014, 03:42:48 AM

@ Pablo

You once mentioned that ButterBot takes about 160 or 165 SMA samples to generate the first EMA value. I have tried various amounts but never seem to be able to replicate the exact same EMA values as butterbot.

Could you ask around about this?

I am compiling a Month to Month Minute accurate excel sheet to distribute on the forum so we can run various kinds of math against the past. But so far I am having a hard time replicating the ButterBot EMA samples. (I'd like it to match the real butter rather than be an approximation).

@ Pablo
Did you get around to investigating the SMA samples?

Puerto,
Email me so we can discuss it in detail

.

Pablo.

Will do.

@ The 1st thresholds.

I'd like to say that if you are creative enough, you can use the first pair of thresholds to limit your losses. Keep that in mind.

If you tune your settings with the first pair, you can also make the bot act slightly different in various types of scenarios.

-----------------

For example, you might want to assure your bot [only] buys after a sustained dip.

In this case your threshold might be a 2 and the second threshold for buying might be a 0.3

Same would be true of a sell scenario.
---------------------

You can also use the first pair to assure the bot weeds out the sudden micro peaks that don't affect price very much. For example, the market might go from -0.18 to 0.3 in two intervals. Using the first pair you can induce the bot to only buy if there is first a 0.25 or higher target and then on the second threshold 0.3

Which means it is looking for progression rather than sudden spikes which only meet 1 criteria.

If you learn how to use the first pair properly you can even set the bot to look for specific types of reversals. Just food for thought if folks think the first pair is useless. It's not. The programmer probably placed it there after realizing that 1 threshold can create wins but also major losses when there are false positives.

If you could set even more thresholds you could probably look for specific types of EMA curves and not just 2 basic thresholds being met.

That is really interesting. I've been trying to find some settings (particularly with the first pair of buy/sell thresholds) that could possibly resist sudden dropdowns like the one happened in mid December. So far without luck.