I'm not entirely sure what method you backtested the theory with, but Librium works precisely because it doesn't care about ups or down, it just equalises the balances every chance it gets. A backtest that makes a loss (or indeed a gain when mining rewards aren't considered) isn't the bot I'm describing, because I invented this bot specifically to reduce/nullify losses.
So far it's working nicely, and modest as the gains are, it's much more than when one is mining and holding all GHS, and less terrifying than letting a more ambitious tradebot handle my cex (and potentially make a loss) while I sleep. In fact during the last hour of cex's charts sloping downwards, it's made a profit. As expected! Approximately the same profit per hour as during the rise last night. As expected!
Here's the account that just holds GHS:
Feb 14 2014 09:18:36 -> GHS 12.8832026400 GHS
--------------------- UPTIME: 12.8880261050 hrs Loop533
----------------- STARTFOLIO: 0.6276280300 BTC total
------------------ MINE RATE: 0.0003780077 BTC / hr
------------------ PORTFOLIO: 0.6324998028 BTC total
--------------------- PROFIT: 0.0048717728 BTC total
and here's the account with hardware miners:
Feb 14 2014 09:18:19 -> GHS 0.5686118200 GHS
--------------------- UPTIME: 15.5889653464 hrs Loop665
----------------- STARTFOLIO: 0.0261674960 BTC total
------------------ MINE RATE: 0.0001128189 BTC / hr
------------------ PORTFOLIO: 0.0279262258 BTC total
--------------------- PROFIT: 0.0017587298 BTC total
It'll get its own thread soon one I've tested it further. Weird to hope for a downswing in the cexgraphs, but that's what I need to stress-test it.