the two answers above are completely correct.
To my knowledge, the most cost-efficient and cheap usb miner is the gekkoscience compaq, which mines @ 8-16Gh @0.3W (best case scenario, without the raspberry pi or laptop or pc you'll need to mine)
If you put the absolute best case scenario in here:
http://www.coinwarz.com/calculators/bitcoin-mining-calculator/?h=16&p=0.3&pc=0.05&pf=0.00&d=178678307671.68800000&r=25.00000000&er=427.40000000&hc=0.00You'll never break even (look at the numbers: days to break even NA)... and this was the absolute best case scenario (actually, not even best case scenario, but a dream scenario that can never exist), assuming following things:
- the blocks didn't halve
- the diff didn't rise
- the price didn't fall
- the stick actually did 16Ghs 24/7/365 and only used 0.3W
- no powered usb hub was needed
- no raspberry pi was needed
- you got all your hardware for free
- no pool fees
- electricity rate @ 5 cents/Kwu
These assumptions are not realistic, and still, you cannot make a profit... Sorry
EDIT: the only way you can actually use this, is as a lotto miner...
I've put some more realistic numbers in the calculator (still not completely realistic, but at least, some more real world, still assuming you got the hardware for free tough)
http://www.coinwarz.com/calculators/bitcoin-mining-calculator/?h=12&p=2&pc=0.10&pf=1.00&d=178678307671.68800000&r=25.00000000&er=427.40000000&hc=0.00This calculation learns us that at current diff, each day at current diff you'll have 1 chance in 740.000 to mine a block... but each week it'll cost you about 3 cents worth of power.
If you setup a cheap raspberry pi + a compaq in your attick, and leave it running, you can see this as a lotto ticket... A onetime fee to buy the hardware, and afterwards you'll have 1 chance in 106.000 a week to mine a block (as long as the diff doesn't rise to much)
At the moment, a block will give you 25
BTC, but in less than 3 months it'll only be 12.5
BTC