The point of arbitrage is that its a zero risk game if done properly so no you cant ruin yourself in a day.
The only way to effectively arb is to keep track of how many BTC/USD shifting you need to do, how reliable is the output from exchange volume/walls however if you keep the shifting small enough there is literally 0 risk. All that happens is buying BTC at low exchange and selling the same amount of BTC at the high exchange thus making $x on the shift and still having exact same amount of BTC afterwards however gaining $ out of the deal.
The problem is that you need to keep ~50% of your balance in bitcoins in order to do these trades. So, for example, in November the bitcoin price dropped around 50% in one day. If you were doing arbitrage without additional security measures, you would've lost 25% of your money (in fiat). I believe that the fact that bitscalper pays in bitcoins and not in USD is also part of the strategy to defend agains such drops.
Regarding 0.6% a day, yes it is acceptable. If I consider arbitrage alone I don't make as much, but bitscalper probably trades in much more markets than I do (and they probably have a better system).
Now thats the flaw in your logic, when you arb the focus is allways on taking the edge from the $ earned on arbitrage between exchanges on the same volume of BTC.
Lets assume you have (to keep it basic and minimalistic) $100 on all 3 exchanges with volume and also 10BTC on each exchange, that is 30BTC that is BTC you never intend to match at a $ rate since the only $ that matters to you is when you swop 1BTC from exchange 1 with 1BTC at exchange 2.
Arbitrage doesnt care if BTC is worth 50c tomorrow or $10 today just aslong as there is another exchange to swop atleast the same amount of BTC to get a $ edge.
You never lose BTC value, you only gain $ value when done properly. Im not sure why bitscalper pays in BTC , the $ approach is much less sophisticated.