Marth's Uthrow Combos on Fox: A Simplified Table (NTSC, 1P)

Fox pre-uthrow %	No DI	Full DI	Slight Toward DI
0-17	Regrab (EASY)	Regrab (EASY)	Regrab (EASY)
18-46 (up until 46)	Pivot regrab (HARD at 16-19) Suggested: Turnaround utilt at 30% to avoid having to do as many pivot grabs (May be SDIable)	Dash Regrab (EASY)	Pivot Regrab (MEDIUM)
47-55	Pivot uair -> regrab (VERY HARD) OR DI mixup: between Delayed turnaround utilt -> middle-FF-timing uair -> Fsmash OR continued juggle (loses to away DI) And middle hit usmash->regrab (grants only a tech chase on toward DI)	Dash Regrab (only on toward DI) dash rising tipper uair -> regrab (MEDIUM)	Soft turnaround rising tipper uair -> regrab (MEDIUM)
56-60 (high 50s)	Rising tipper uair -> rising tipper uair or Fsmash -> Fsmash (HARD)	Dash rising tipper uair -> regrab (MEDIUM)	Soft turnaround rising tipper uair -> regrab (MEDIUM)
61-64 (low 60s)	Rising tipper uair -> Fsmash (HARD)	Dash rising tipper uair -> regrab or Fsmash if possible (MEDIUM)	Dash dance rising tipper uair -> Fsmash (MEDIUM)
65-74 (high 60s and low 70s)	Middle-timing-FF soft uair (delay as much as height and % allow) -> regrab (MEDIUM) Update: If soft uair is DI'd toward, Fsmash, pivot if needed	Dash rising tipper uair (delay as much as height and % allow) -> Fsmash (HARD)	Dash dance rising tipper uair -> Fsmash (HARD)
75-79 (high 70s)	Middle-timing-FF soft uair (delay as much as height and % allow) -> regrab (MEDIUM) Update: If soft uair is DI'd toward, Fsmash, pivot if needed	Dash rising tipper uair (delay as much as height and % allow) -> Fsmash (HARD)	SH back, Middle-timing-FF soft uair (delay as much as height and % allow) -> regrab (MEDIUM)
80-85 (low 80s)	Fsmash (EASY-HARD)	Pivot Fsmash (MEDIUM)	(walk) Fsmash (EASY-HARD)
80-94 (80s to low 90s)	Falling soft uair -> Fsmash (pivot if needed) (HARD)	DI mixup: between fair and up-B (MEDIUM to HARD) OR Pivot Fsmash ((VERY) HARD)	SH back, Falling soft uair -> Fsmash (HARD)
95-110 (high 90s to 100s)	Nair (EASY) OR DI mixup: between up-B and a choice of either soft/reverse fair/bair juggle or SH back side-B -> uair -> up-B, depending on what side you want to send them	Nair OR DI mixup: between fair and dash SH side-B -> up-B	Nair (EASY) OR DI mixup: between up-B and a choice of either soft/reverse fair/bair juggle or SH side-B -> uair -> up-B, depending on what side you want to send them
110-120 (110s)	Nair (VERY EASY)	Fair (EASY)	Tipper Bair (EASY)
121-140	Nair (EASY)	nothing	Tipper Bair (EASY)
141-170	FJ DJ Uair (EASY)	nothing	Bair up to 155 (EASY)

Update: I had the chaingrab % ranges wrong.  The chaingrab on no DI ends at 46% with lower port number (52 for away, 61 for toward).  This makes it even more difficult to get through to later percents without a mixup, as pivot uair is very hard at 47%.  Consider saving your pummel or going for a double pummel at this percent.  Or, just take the utilt mixup.

The table tells you what to do at each percent for each DI.  The percents listed refer to the percent that the opponent has before you throw them, so you can glance at it once you have the grab, pummel if necessary, and narrow down what options you're going to execute depending on the DI.  It assumes you are playing on Final Destination, but is a solid base to build off of for combos anywhere else.

Yet another combo guide?

So, nothing mind-blowing, right? Hasn't this been done before? Didn't M2K settle this in like 2008? No, not quite.  Every combo guide I've seen so far has been a hodge podge of percent ranges at which options work and suggestions as to what to do at each percent, often with bad or incomplete advice. The information given is inconsistent across guides, even ones claiming to give exact percent ranges at which options work.

My table is no such guide--I have designed it to be easy to remember and execute in real matches, without losing any effectiveness or listing too many superfluous options.  The percent ranges given are not the endpoints of possibility for the followups listed for the reason that you should not be trying to execute them at those ranges: there's plenty of overlap between the ranges of possibility for the various followup options.  At the endpoints, you approach one-frame links, as those are the points where the move barely connects.  Why subject yourself to that when you usually have an easier alternative? Beyond being imperfect, you need some leniency in order to handle SDI or strange DI choices and give yourself time to react.  So, I strategically chose percent ranges that would make the followups easy to do, the success of which is shown in the difficulty levels given.   They also align the ranges that would normally differ wildly for each kind of DI so that they fit nicely into a table and so that the combo can be divided into only handful of parts rather than an impossible-to-remember list of percent ranges for everything.   The overlap of possibility ranges means that the given ranges are not ironclad; you can choose to do a followup from a neighboring cell in the column if you're at the borderline in the given ranges.  The followups in the later percents push you closer to the limit, as there is a reward for doing so: finishing the combo at the highest percent possible.  At the very least, you can be sure that everything in the chart actually works and does not approach impracticality.  

I also prioritize guaranteed followups over mixups or gimmicks.  A mixup is strategy in which your force your opponent to choose between a small number of options, and a gimmick is a strategy which only works if your opponent is not prepared.  There are places for these types of strategies, but taking the guaranteed followup is often the better choice because it guarantees a payoff.  In a two-choice mixup, the expected payoff is the average of the reward of the two options cut in half, assuming the two options of both players are done equally as often.  A gimmicky strategy may or may not work, depending on your opponent, you are just hoping that it will.  Everything in this table will definitely work if you execute it properly and follow their DI.  Further, I only list one guaranteed followup unless there are other alternatives that aren't strictly worse.  This isn't a fashion show, just do the best followup.  You can be creative if something goes wrong.

The followups listed are short combos for a reason.  DI and SDI cause there to be a lot of variance in long combos.  It may be easy to follow DI with regrabs, but the combo enders, tipper fsmashes, are much more difficult to line up in chaotic situations, which is the most important part. Limiting combos to two to three hits after the throw limits the number of different situations you'll have to face, allowing you the repetition necessary for mastery.  

Above all, it's more important to know what you should do and be prepared to do it rather than know everything else.  That is what this guide attempts to provide.

The Flowcharts

The table tells you the best followups for each percent, but a flowchart best represents full combos which involve many uthrows.  This simple flowchart is a reduction of the table followups to a combo strategy that nets the highest minimum reward for landing an uthrow below 95%.

With this, we can see how much I've reduced the combo tree into a simple, elegant solution. Let's get into the reasoning for some of the choices I've made in the inclusion and exclusion of some followups in the table and flowchart. There's a few popular moves that I avoid listing: utilt and fair.  These moves are usually part of a two-choice mixup that work depending on their DI.  Utilt is especially volatile due to the number of different hitboxes it might hit with, which have varying knockback and trajectory, but generally the opponent will escape the combo if they DI in the opposite direction that Marth is facing (beyond a low percent)  The latest hit of utilt has the most vertical trajectory, so if you choose to use it all, hit with the back part.  Fair is similar, they only need to DI away and they will escape.  Less popular is usmash, which does find a place in my table as part of an alternative mixup option, as it is weak to DI toward.  The usmash/utilt mixup is not shown in the simple flowchart, as it assumes that you are able to execute the superior followup, pivot uair.

Instead we see a lot of regrabs, uairs, and Fsmashes--all moves that are good no matter what the DI is.  If you execute the flowchart properly, you should end with an Fsmash with a post-hit percentage in the 90s-110s, which is sometimes enough to kill outright, but at least nets a strong edgeguard situation.  Being able to pivot Fsmash is required to follow DI properly in some cases of uair to Fsmash.

The pivot uair at 50-55% on no DI and the pivot Fsmash at the later percents, so the table includes alternate options.  They are mixups, so they will only work half the time if you don't have a read on your opponent, but they are preferable if you cannot land the guaranteed followup over the half the time.  Here's what the flowchart looks like with mixups replacing the difficult followups:

It's looks a bit more complicated, but it's only a rearrangement of the grab states to match the DI directions and the insertion of the mixup diamonds, which have four transitions, a success and a failure for each option.

Both flowcharts neglect the details of the chaingrab portion of the combo.  Reference the table for when you need to dash or pivot grab.  The table and the flowcharts don't take stage position into account, but if you can land a dair that will send them offstage, go for it (it works from about 23-115% on side DI).

Not Always Enough

That would be all there is to it, but there's one problem: the opponent can strategically DI in order to take the weakest ender at favorable stage positioning.  Unless some changes to the flowchart are made, no matter where you get the initial grab, the earliest point at which you have a strong ender is determined by the opponent's DI choice.  Every time one of the grab states is reached, the opponent has the opportunity to DI toward center stage, reversing any progress you made toward push them to the edge.  The optimal opponent will force the combo to flow into "Lowest Fsmash Ender" to the longer side of the stage, what I call a "cross-stage Fsmash."

The crude diagram shows a potential scenario where the opponent DIs toward the edge at the 50-55% mark when he is relatively safe in doing so, then DIs toward center stage and survival DIs. After a cross-stage Fsmash, the opponent may easily survive by using survival DI and guessing right on the edgeguard.  In a severe enough cross-stage situation where you're forced to end the combo, not even a fair will help you much at this point.  A fair will beat their survival DI, sure, but it's a long way to continue that aerial carry combo into a dair.  Some careful planning must be done to ensure that you avoid severe cross-stage ender scenarios and that when you do choose to do an aerial carry combo, you know the proper setups and mixups to employ at various percents.

Let's start with some general rules.  First, know that every fair is part of a mixup, where if they DI toward or do survival DI, you will guarantee a dair or another aerial, sometimes an Fsmash (SH fairs will net an Fsmash at around the same percentages where SH uairs will).  If they DI away, they escape, but it may net a weak edgeguard anyway, especially in places where your Ken combo got denied.

The other part of the mixup are moves that you were already doing: Fsmashes and uairs.  Fsmashes are not always possible, but if they DI away, in fear of a fair, they may die at hilariously low percents to a DI'd away Fsmash.  In the places where I include Fsmash in the table and flowchart, they are usually great no matter how they DI. Uairs often net a followup no matter where they DI, which is why it's included so often in my guaranteed-followup-centric guide, but they DI toward on your uair, they will end up on the other side of you.  This won't mess up your followup if you fast fall and L-cancel the uair as opposed to auto-cancelling, but it will reverse the stage orientation, which can cause a cross-stage situation.

Fairs are only needed to plug the few holes in the regrab/uair/Fsmash-heavy combo strategy by stopping stage position reversals.  Use it opportunistically, when you find yourself in a good stage position that you want to keep.  Usually, you can do just fine by executing the strategy laid out so far and edgeguarding effectively.  Avoid fair when you can, for the previously stated reason that you should avoid mixups unless the risk/reward is worth it.  Remember, when you go for a mixup, you are flipping a coin on the success of your combo.  (Pivot) Fsmash is the alternative way to maintain stage positioning. If the combo tree resulting from landing the fair and guessing right on their DI doesn't improve on the alternative that much, it's not worth it!

Now let's look at how fair combo trees actually play out.  Just jumping in and waving around your sword without plan is not necessarily going to net you that sweet Ken combo, even you do guess right on their DI.  Each combo is a series of mixups, whose parts are separated by a '/'. Each mixup is a coin flip on continuing the combo or ending it well, otherwise, the mixup would not be threatening enough on both sides and the opponent would have a safe option to use when in doubt.  Tipper hit everything actually, tipper hit everything at low percents, while either soft fair or tipper fair work at high-mid percents (~80%) while you should do soft fairs after about 100.  It really depends on how high up they are, whether you're SHing or FHing and whether they're doing full toward or survival DI.  It'd be very difficult to test when to do what thoroughly, so I can only give general rules.  Soft fairs give you another knockback angle to work with. The goal with fairs is to beat toward DI but also maintain stage orientation or carry them across the stage, and soft fair accomplishes the latter more effectively, but it can't be used at low to mid percents due to the lack of hitstun and knockback and how limited your options are even if they DI toward.  Doing soft fair as the last part of the mixup is useful since if they DI away, the knockback of soft fair can put them into an edgeguardable position.    Always SH for uairs.  You can always just Fsmash at the beginning as part of the mixup.  All listed followups can be replaced by a dair and the end followup can be replaced with a soft fair or up-B if you don't want to spike them into the ground, though don't bother with these unless you're going for the dair or are using your opponent's fear of the Ken combo.  These are all rough percent ranges.

Stage carry combos (side-DI)

53-57: SH rising fair / Fsmash, double aerial: fair/uair (->fsmash), SH fair/uair (->dair)

58-63: Dash Shield stop SH fair, ???
64-69: SH rising fair FF / SH uair(->fsmash), SH or FH fair / SH uair (->dair)

70-79: SH fair / uair (->fsmash/dair)

80-85: SH or FH fair / uair (-> dair)

86-94: FH fair / pivot fsmash or up-B

• At 53-57 it's questionable whether this combo tree survives SDI in the early part.  The first mixup is between a low % Fsmash and the beginning of a SH double aerial.  The Fsmash will be enough in some cases where they DI away for some reason.  They're pretty safe to DI toward here, as you have you guess right twice before gaining much from this setup: first on the fair, then on the second part of the SH double aerial, either fair or uair (or dair).  Guessing wrong on the first fair leads to a tech chase, which might not be so bad by the corner if they don't slip off.  Guessing wrong on the second fair ends the combo but guessing wrong on the uair (you always want them to DI away against uair) puts them in a workable place to continue the juggle, albeit on the wrong side.  Guessing right on both aerials puts them right in Fsmash tipper range, where you can also choose to go for a 3rd or even 4th guess.  Have fun completing this one 1/8th or 1/16th of the time, when the stars align.

• At 58-63 I recommend not going for it.  The first SH fair sends them too high to do a double aerial, yet not high enough for you to land and tipper Fsmash.  You could do a second SH fair, but it's awkward to line up, you need to do a shield stop to control your momentum for the first one.  Didn't even bother completing this one.  This is why this analysis is useful, so you know when to avoid trying for a fair.

• 64-69% is the ideal percent range to go for this, and luckily, this is right where you end up after a 50-55% uair->regrab.  Here, you have a credible threat of an Fsmash on the first and second mixup.  You have plenty of time to line up your tipper fairs.  The percentage is high enough that you get a reasonable amount of stage carry but not too high that it's not worth it to just Fsmash.  You might be able to get a third mixup, but it'll involve an up-B.  (up-B/fair (->dair))

• At 70-79% they may be too high to Fsmash after the fair.  They get carried a good distance on an uair.  They're already a pretty high percent, though, so you may want to just Fsmash here if the usual uair->Fsmash isn't the better option whether they DI toward or not.  Just know that they're generally too high % to continue the combo with a threatening mixup any further than just one mixup.

• At 80-85 and 86-94 they high enough that it probably isn't worth it, but they do get carried further than usual with a single guess because their % is so high and the dair will be higher up, increasing its ability to clear the edge.  It is a good alternative if you aren't confident in your pivot Fsmash.

So the whole point of this last section was to add in the option to go for a fair combo at 64-69%. That's the only time when it definitely seems worth it to either diminish the effect of a cross stage situation or to avoid one entirely using a fair combo.  If you use utilt at 50-55%, that's also a good place to convert into a stage carry combo, and in rare cases I'd go for it at 53-57 or just Fsmash at 58-63 if uair->regrab would put us in a really bad spot.

Conclusion

So, there's almost everything that needs to be said about Marth on Fox uthrow combos.  It's not enough to just read this post, though, readers should practice each followup to be able to execute them and internalize the strategy.  Let me know if you found this useful or if I got anything wrong.  Just look at the flow charts if you don't want to read all of this.

Don't worry, I didn't forget Falco, like everyone else.  A table for him is complete, and I will do his flowchart later.  A Falcon table is also complete, but it doesn't make much sense to make a flowchart for him.

Additional notes:

• This is a guide for NTSC, as opposed to PAL version, the one used in Europe and Australia.  Fox is a bit lighter in those versions, changing the combo in unpredictable ways.
• This is all tested using the first player port.  If you have a higher player port than your opponent, then the opponent will start their throw hitstun a frame later than usual, effectively giving you an extra frame of leniency without changing the timing your attack (this is retarded).  This doesn't affect the combo guide much because it rarely approaches one frame links where the single extra makes a big difference.  In general, you should try to get the higher port, though, as it will make the chaingrab slightly easier.
• There are multiple degrees of slight DI, as many as there are DI directions.  The leniency of the percent ranges in this guide make that variation not matter much.  The particular slight DI I used in the testing was the one that the AI does in Training Mode when you set it to Attack.
• I'm eventually going to include a link to an explanation in each cell in the table.
• This guide doesn't include all the nuance of the chaingrab part of the combo nor does it explore the options for the later percents (using a different throw might be better)
• When I set out to create this, it was supposed to be a beginner's easy and simplified version of the combo tree, but it ended up being useful for everybody, as the guide is a strategy that makes the combo easier for everybody, and I offer easy and difficult alternatives in some cases.  In any case, it's a solid base to build of off and there only seem to be minor optimizations that can be made here and there, if any.