UPDATE: As of MongoDB 2.4 one would use a "text" index and full text search query to do this. You can read about them here. If using a recent MongoDB the approach below would be silly and unecessary.
However, if you have MongoDB < 2.4.0 you could use a regular expression like so:
> db.reg.insert({searchword: "win"})
> db.reg.insert({searchword: "window"})
> db.reg.insert({searchword: "Win"})
> db.reg.find()
{ "_id" : ObjectId("4ecd2e33dd68c9021e453d12"), "searchword" : "win" }
{ "_id" : ObjectId("4ecd2e36dd68c9021e453d13"), "searchword" : "window" }
{ "_id" : ObjectId("4ecd2e39dd68c9021e453d14"), "searchword" : "Win" }
> db.reg.find({ searchword: /^win$/i })
{ "_id" : ObjectId("4ecd2e33dd68c9021e453d12"), "searchword" : "win" }
{ "_id" : ObjectId("4ecd2e39dd68c9021e453d14"), "searchword" : "Win" }
However, your version wasn't working because you don't need the "/"s when using the $regex operator:
> db.reg.find({ searchword: { $regex: "^win$", $options: '-i' }})
{ "_id" : ObjectId("4ecd2e33dd68c9021e453d12"), "searchword" : "win" }
{ "_id" : ObjectId("4ecd2e39dd68c9021e453d14"), "searchword" : "Win" }
Please note that case insensitive queries do not use the index so it might make sense to make a lowercase searchword field so that you can speed that query up.
Go here for more info on RegularExpressions