I have tried every version of Margin on my MacBook Pro as they come out and every single one has crashed after running one or two days. Support appears to be unable to correct it. And every time Margin crashes I go back to the last version of Leonardo, which is ROCK SOLID, reconfigure 14 pairs on Bollinger Band and start over. At what point will I give up entirely on Margin and just stay with Leonardo?
I have the exact same problem on my laptop (High Sierra 10.13.4) and haven't been able to narrow it down. Margin only runs for about 24-72 hours before it "quits unexpectedly." At first I thought it was triggered by having too many exchanges and pairs (running Binance and Bittrex), but even after reducing the footprint to just Binance, with less than the max recommended pairs, and bare charts without any added indicators, the problem persists.
I'm considering either setting up a Linux VM or running it on my Windows PC to see if it's OS related, if I find the patience.
There's a tradeoff between the two applications. What I've noticed is that Margin is relatively light on memory usage, but destabilizes within a couple days. Leonardo eats a huge amount of RAM/swap (possibly a slow leak) but is otherwise rock solid over literally several weeks. My last instance set a personal record at 52 days of uptime before I shut down to update the OS, though by then it showed >100GB memory usage in Activity Monitor and had to be force quit. This was again with Binance plus Bittrex and around the max number of recommended pairs.
New bots, drawing tools, and social media integration are useless to me if I have to babysit the application, and more complexity will just make it harder to narrow down stability issues. I can try to keep troubleshooting to see if there's a pattern with the crashes, but the team should be able to easily test this scenario. Use the latest versions of macOS and Margin, open Binance and Bittrex, add the max recommended number of pairs to each, and just monitor it for a few days. That would at least tell us if the error is reproducible on your end or if it's something specific to our setups.