Hi
new to both Quantopian and Python so very much a noob trying to become familiar with the ecosystem, and as such i have some questions:

(1) It took me a while, but i have started to understand the point of alpha-lens with tear-sheets is to be able to test factors without needing a lengthy backtest. So i play with some simple factors as practice, and i don't pretend to understand everything in the tear-sheet results, but i think a positive IC with a low p-value is 'good'? (as in... don't throw this factor away yet). Is that fair?

(2) However, when i move the factor to a backtest it is terrible. Really bad. Is this a common occurrence? Or an indication that i have done something wrong in translating the factor to an algo?

eg for some factor trying to catch short term moves tested over the QTradableStocksUS the tear-sheet shows
1-day period: ann. alpha = 0.058, IC=0.01, p-value=0.002
5-day period IC is double and p-value is smaller (but this is suspicious given the general tendency of the market to drift up i'm not sure if this is accounted for)?

Over same period the backtest is horrible. Does this happen often? Was my IC not high enough to keep looking?
If not, are there any rules of thumb for reasonable tear-sheet values to move to backtest?

PS: does ann. alpha basically mean CAGR (compound annual growth rate)?

(3) if i'm only going long, how can i get -1000% return? Doesn't my cash run out at -100%?
I start to realise maybe leverage is involved somehow, but the api docs for 'Ordering' under 'Algorithm Reference' don't mention it!?!
How can i tell the backtester not to buy anything if i don't have any cash?
Or if that's a calculation i am supposed to do myself, how can i take slippage/commission into account when doing so?

I actually have so many more questions, but i guess they're my top ones for now!
Would appreciate any guidance possible. Thanks v. much.